YHMAN Shared Virtual Data Centre

Business Case

Final

Document Author – Ted Edmondson, Executive Advisor, KPMG LLP Document Reviewer – Adam Stork, Director, KPMG LLP

Final

YHMAN Shared Virtual Data Centre

Business Case Final

Contents

1

Executive Overview

5

2

Introduction

9

2.1 2.2 2.3 2.4 2.5 2.6 2.7 2.8 2.9 2.10 2.11

Purpose of the Document Status of the Business Case YHMAN Background SVDC Background SVDC Vision SVDC Strategic Aims SVDC Delivery Scenarios SVDC Market Differentiator University Data Centre Regional Analysis Market Analysis Legal Constraints

9 9 9 10 11 11 11 14 15 16 17

3

Strategic Fit

18

3.1 3.2 3.3 3.4 3.5 3.6 3.7 3.8 3.9

HEFCE Goals Alignment of SVDC to HEFCE Replicability Alignment of SVDC to the Joint Venture Universities Benefits Strategic Benefits Strategic Success Factors Strategic Risks and Mitigations Governance and Stakeholders

18 19 19 20 20 22 22 23 23

4

Delivery Options Analysis

24

4.1 4.2 4.3 4.4 4.5 4.6

Step One – Requirement Scenarios and Delivery Options Step Two – Indicative Costs of Sourcing DC space Step Three – Indicative costs and benefits Virtualisation by Universities Virtualisation by YHMAN Recommendations

25 27 30 36 39 41

5

Commercial Considerations

48

5.1 5.2

Costs and Affordability Consequences of Not Proceeding

48 48

6

Corporate Structure Options appraisal

50

6.1 6.2

Summary Corporate Structure Options Appraisal

50 50

2

YHMAN Shared Virtual Data Centre

Business Case Final

6.3 6.4 6.5 6.6

Advantages and Disadvantages of Corporate Structure Models Considered Implications of VAT on the SVDC Facilities Management Arrangements Agreement on Financing of Shared Services Vehicle

51 53 57 57

7

Conclusion

58

8

Appendix I – Costs of Data Centre provision

59

@ 2009 KPMG LLP, the UK member firm of KPMG International, a Swiss cooperative. All rights reserved. KPMG and the KPMG logo are registered trademarks of KPMG International.

3

YHMAN Shared Virtual Data Centre

Business Case Final

Disclaimer This Business Case has been prepared on the basis set out in the Contract signed by Ed Carter on behalf of YHMAN Ltd (known as the Client), dated 15 December 2008, and should be read in conjunction with our original proposal dated 1 December 2008, to which that contract refers. This Report is for the benefit of only the Client, and has been released to the Client on the basis that it shall not be copied, referred to or disclosed, in whole or in part, without our prior written consent. Nothing in this Report constitutes a valuation or legal advice. We have not verified the reliability or accuracy of any information obtained in the course of our work, other than to ask the client to confirm that they are content with its usage. This Report is not suitable to be relied on by any party wishing to acquire rights against KPMG LLP (other than the Beneficiaries) for any purpose or in any context. Any party other than the Beneficiaries that obtains access to this Report or a copy (under the Freedom of Information Act 2000 or otherwise) and chooses to rely on this Report (or any part of it) does so at its own risk. To the fullest extent permitted by law, KPMG LLP does not assume any responsibility and will not accept any liability in respect of this Report to any party other than the Client.

4

YHMAN Shared Virtual Data Centre

Business Case Final

1

Executive Overview Yorkshire & Humberside Metropolitan Academic Network (YHMAN) currently provides networking services to eight Joint Venture universities across the Yorkshire & Humberside region: Leeds, Leeds Metropolitan, York, Bradford, Huddersfield, Sheffield, Sheffield Hallam and Hull. Portal technologies, collaboration tools, multimedia applications, back office automation packages and many other applications all require increased access to computing power and storage to work effectively. These requirements for greater power and storage are increased further by greater expectations from end users that services will be always on, highly available and accessible from anywhere with an internet connection. This increase in requirements for infrastructure is further increased by a requirement to provide suitable Business Continuity and Disaster Recovery (BCDR) facilities. Accordingly some of the Joint Venture universities are experiencing pressure on their data centres (DCs) due to increased demand for services, and others have just brought, or are about to bring extra DC capacity on stream. None of the Joint Venture universities have access to true cross-city BCDR facilities, indeed the potential insufficiency of some universities BCDR arrangement was highlighted during recent flooding in the Yorkshire and Humberside area. With this in mind YHMAN is seeking to investigate the potential for utilising the current or imminent ‘headroom’ in some university DCs to meet the increased demand in others, as well as providing true cross-city BCDR for its members. There is an increasing recognition that shared services in Higher Education (HE) could offer potentially significant cost savings for universities. With this in mind, and the continuing success of YHMAN as an organisation, jointly owned by eight universities, the Board of YHMAN implemented the Shared Virtual Data Centre (SVDC) project to investigate the feasibility of sharing DC services. At its simplest, the service offered by the SVDC would be DC rack space in which to house servers. Taking the service offering one level higher, the SVDC could offer physical DC infrastructure. Most organisations who are doing this have quickly abstracted to the next level, which is offering access to virtual infrastructure. In this instance, universities would pay a flat rate for access to a certain number of servers (physical or virtual) and storage. This makes more efficient use of the physical infrastructure and provides better resilience. Potentially other services could also be offered including a virtualised storage platform, shared application provision and support, shared commodity business processing, and joint service offerings such as shared email, finance systems etc.

This report was signed off at the YHMAN Board 11th June 2009 with all eight joint venture universities committeed to the SVDC implementation subject to successful completion of the recommended technical proof of concept and pathfinder project. Purpose of the document

5

YHMAN Shared Virtual Data Centre

Business Case Final

This document is not intended to be a very detailed business case that examines the benefits and risks at a very detailed level and provides definitive costs for moving forward with a preferred option. Rather, it examines the issues at a high level, uses assumption based planning to look at potential benefits and makes a high level recommendation on whether it is worth proceeding to the next stage. At a high level, it examines the sharing of DC space and virtual infrastructure. It does not cover other services, since these would need to be tested individually with a relevant stakeholder groups to ascertain political acceptability, and be assessed from a technical and business process perspective to ensure that these are feasible services to share. We believe there is a valid business case for sharing services more broadly across the HE sector, however this is beyond the scope of this document. Costs and affordability of the SVDC project vary according to how many universities collaborate to provide the service via YHMAN, our analysis shows that for one or two universities the benefits would be marginal, however for five universities, with a combined requirement for 200 racks of DC space, the combined financial benefits could be as much as £4m over a five year period. If virtualisation is undertaken by the individual universities, that benefit could rise to a collective £5.5m over the same five year period. Furthermore if virtualisation is undertaken by YHMAN centrally, and YHMAN offers computing capacity to the universities as a service, the corresponding benefits could increase to £10.7m. There remain two questions from this analysis: Firstly, whether the shared services can provide acceptable service levels to the Universities? This report recommends a pathfinder exercise to initiate the service provision and demonstrate that the technical, process, and cultural changes are possible to enable a high level of service provision. Secondly, whether the price point for the services would be attractive to the Universities? Our analysis demonstrates, on a scenario basis, that the service should prove attractive if at least two universities sign up for the services. The analysis also shows that the savings increase for each additional participating university. The report recommends that a joint planning exercise is carried out across the universities to enable them to decide to sign-up for the service, and to enable YHMAN to plan the optimum level of provision over the coming years. That joint planning should be carried out in parallel to the pathfinder. One further consideration concerns the current availability of ‘spare’ racks around the region to allow YHMAN to lease them. Currently, the Joint Venture universities have indicated that there is very little spare provision, meaning some injection of spare capacity is required. One way to facilitate this would be to construct a YHMAN data centre, into which universities could move some of their infrastructure. That would create spare capacity that YHMAN could then invest in to bring it up to an agreed standard, and expand their service offering. The report concludes with a summary of the potential for YHMAN to deliver further shared IT service provision beyond DC services. This potential should be investigated in detail once a robust core DC service is being delivered.

6

YHMAN Shared Virtual Data Centre

Business Case Final

Recommendation 1 – Move forward with implementation of the service Due to the benefits we believe could be realised through this project, we recommend that the service be initiated with a small scale pathfinder project that will provide insight into the process, cultural and technical issues thrown up by offering shared DC services in this area. The proposed pathfinder project involves using legacy equipment in one or two JV universities to offer a virtual service to others. The ‘customer’ universities would receive a secure, discrete service of remote access to a number of virtual servers that they could administer. The project will not be focused on providing cost efficiencies or medium to long term financial gain. Rather it will focus on demonstrating successful resolution of the very real and immediate process, service level, and people issues that need to be resolved to move forward with a wider implementation. The Technical Proof of Concept element of this phase can begin early. It will require two universities with enough technical experience to set up and run a small scale proof of concept. Currently Leeds and Sheffield Hallam are well placed to fulfil these roles, subject to the agreement of the YHMAN Board. Recommendation 2 – Plan across the Universities In parallel with the pathfinder project described above, it is recommended that that the Joint Venture universities and YHMAN carry out an extensive joint planning exercise to understand the DC capacity requirements of the Joint Venture universities, build detailed cost estimates for YHMAN provided DC space and hardware, develop service model for the proposed YHMAN services and plan implementation and transition to the live service. This exercise will also be used to agree an ongoing minimum specification for YHMAN provided Data Centre space that all Joint Venture Universities are happy to sign up to, and agree an approach to any contractual or service issues that could arise from data and services being managed and run off the university campus by a separate third party. We recommend that the Joint Strategic Planning is implemented at the same time as the Pathfinder to achieve maximum benefit from both phases of the project. External support to facilitate both recommendations To allow both recommendations to be implemented effectively will require external support for YHMAN and the Joint Venture universities to derive the value they expect from the Pathfinder and the Joint Planning. This support will provide assistance to the Pathfinder in terms of governance, support and assurance that the programme is being run appropriately. The definition of service levels against industry standards will also be part of the external support’s role. For the Joint Planning, it will provide structured approaches to universities working together, and bring in external experience in facilitating such joint working and focussing the discussions on relevant areas of mutual benefit. The external support will also maintain a focus on achievable outputs and drive progress towards agreement on the key benefit areas.

7

YHMAN Shared Virtual Data Centre

Business Case Final

Funding The total Pathfinder and Joint Strategic planning project costs are estimated to be approximately £215,000. YHMAN will seek funding of £150,000 and if they are successful, will match that with £65,000 from their reserves thus demonstrating the commitment of YHMAN to the project. (Note all figures in this document are exclusive of VAT).

8

YHMAN Shared Virtual Data Centre

Business Case Final

2

Introduction The idea of shared services has been investigated and promoted by the UK government for several years, most notably in Central and Local Government where it is seen to be a good way of introducing efficiencies and taking advantage of economies of scale to generate savings. More recently the same ideas have been extended to the education sector, particularly HE. HEFCE is actively seeking ideas and business cases for the implementation of shared services across the HE sector. This business case has been produced on behalf of the Yorkshire and Humberside Metropolitan Academic Network (YHMAN) by KPMG LLP as part of a funding request to HEFCE to provide pilot funding for the Shared Virtual Data Centre (SVDC) concept. The costs and numerical appraisal of the options are based on a number of assumptions and inputs obtained from and agreed with YHMAN and HEFCE. These are presented in more detail in Sections 4. This section of the business plan describes the high level purpose of the document, its status and the background to the work that produced it.

2.1

Purpose of the Document This document is a business case. It provides a strategic overview and commentary on the options available to HEFCE and YHMAN on piloting the SVDC concept and subsequently implementing it more fully. It is intended to enable HEFCE to:

2.2

•

Understand the current market for data centre services in the region

•

Consider an assessment of the preferred option for delivering the SVDC service

•

Review the proposed operating model, corporate structure and governance model to deliver the SVDC service

•

Make a considered decision on the funding it should set aside for the SVDC pilot.

Status of the Business Case This is a draft business case for consideration and comment

2.3

YHMAN Background YHMAN has been providing effective and efficient networking capacity to its eight shareholder universities and other education clients for a number of years. It owns dark fibre across the region and can relatively easily and quickly increase bandwidth to its clients. The network has plenty of spare capacity, and the anticipated growth in bandwidth requirements across the region can easily be accommodated in the current infrastructure. This success in providing networking services has earned YHMAN a level of trust from its shareholder universities that a shared service approach via a jointly owned organisation can work.

9

YHMAN Shared Virtual Data Centre

Business Case Final

The Board of YHMAN is comprised mainly of senior IT managers from the shareholder universities. This inevitably leads to sharing of other information besides the joint interest in YHMAN. Recently the topic of data centres has been relevant to many of the board members as they have been investigating upgrading their data centres or building completely new facilities. Since the floods of 2007 that afflicted whole cities in the region (notably Sheffield) greater attention has also been paid to BCDR. Currently the BCDR provision in the shareholder universities is provided by a secondary data centre. However these secondary data centres are rarely sufficiently distant from the primary facility to provide effective BCDR. Effectively if a disaster threatens the primary facility it often threatens the backup facility too. This happened to one local university during flooding in 2008. The power grid to the university was affected and they had to deal with random outages for four days. The contracts for delivery of diesel to keep the generators running were rendered ineffective by the collective need of public services across the region, leaving them potentially exposed to loss of power to their IT infrastructure. As the secondary facility is connected to the same local power grid as the primary facility, it would have been equally affected by loss of power. The university could have been in the position of being unable to offer IT services to its students who were still attending lectures. With a collective need to upgrade and expand existing data centres, and relocate secondary facilities further away from primary ones the Board of YHMAN decided to investigate whether there was a collective solution that would save them money in the longer term. This brought about the concept of the Shared Virtual Data Centre (SVDC). This idea closely parallels the work being investigated by HEFCE into the Shared High End Data Centre (SHED) project that is looking at the potential for more universities to share fewer data centre facilities across the country. KPMG has been working with YHMAN and HEFCE to assess the feasibility of the SVDC and produce this business case since December 2008. The overall objective of this work was to produce a business case to present to HEFCE and the shareholder universities with a view to securing funding to allow a pathfinder project to be undertaken. This would provide the foundation for a wider implementation of the SVDC across the region, and the business case will address both the pathfinder and wider implementations to assess the overall viability of the project.

2.4

SVDC Background Since the inception of data centres as a central facility to house the entire central computing infrastructure of an organisation, there has been an inexorable rise in the space requirements for such centres. This is as true in the HE sector as any other. As more and more aspects of university life are conducted via the medium of ICT so the number of applications in use has grown. This growth in applications leads to a concomitant growth in the computing power needed to run those applications and the storage capacity for the corresponding data. Furthermore e-mail, student information systems, CRM systems, VLEs, shared folders and online presentation of research and teaching materials have all contributed to increased storage requirements.

10

YHMAN Shared Virtual Data Centre

Business Case Final

As server and storage requirements have increased, so has pressure on existing data centre space. Several of the Joint Venture universities are at or reaching capacity in their existing facilities and need to be able to access extra capacity in the short to medium term. Alongside this need for extra space is the increasing awareness of the importance of BCDR to a university. With this in mind YHMAN is seeking to investigate the potential for utilising the current or imminent ‘headroom’ in some university DCs to meet the increased demand in others, as well as providing true cross-city BCDR for its members.

2.5

SVDC Vision The SVDC vision is to maximise the use of existing and planned data centre capacity across the Yorkshire & Humberside HE community and facilitate improved levels of BCDR provision ultimately leading to region wide strategic planning of data centre capacity requirements. To achieve this we aim to work closely with all HE institutions across the region, leasing space from and to these institutions to make best use of the capacity available and providing BCDR options that increase the resilience of these institutions’ existing IT infrastructure.

2.6

SVDC Strategic Aims 1. To bring together the Joint Venture universities, YHMAN and HEFCE to asses the feasibility of sharing DC services and virtual capacity 2. To transition from an independently sourced university DC model to a centrally sourced YHMAN DC model via an initial small-scale pilot

2.7

SVDC Delivery Scenarios

2.7.1

Technical Proof of Concept and Pathfinder The investigation into the feasibility of the proposed SVDC will be run in two phases: the Technical Proof of Concept (TPoC) and the Pathfinder itself. The TPoC will focus solely on investigating the technical feasibility of managing a virtual service across a network. It will be run from legacy equipment provided by one or two of the Joint Venture universities and hosted in one of those universities. It will require a second university to manage the ‘other end’ of the service to provide a controlled scenario in which to investigate the technical issues that may arise. Sheffield Hallam university and Leeds university have already put the infrastructure in place for the TPoC and have begun technical trials. The TPoC can be running fully extremely quickly, as much of the infrastructure requirements and configurations are already in place. It is recommended that this is enacted quickly so that the technical learning can be gained while funding is pursued for the main Pathfinder.

11

YHMAN Shared Virtual Data Centre

Business Case Final

During the pathfinder project, YHMAN will provide a limited data centre service to its customers. This will require two ‘host’ universities to be identified. The Pathfinder infrastructure will be installed and hosted in these two universities and will provide virtual server infrastructure to the client universities. These virtual servers will be configured for client universities to access remotely, and will prove the concept of remote provision of service, and virtualisation. No firm decisions have been taken on the participating universities as yet, nor the client universities. Several are viable choices for either role, but discussions need to be continued and terms of reference agreed before any final decisions are made. It is recognised that commitment will be required prior to requesting any funding from HEFCE or the YHMAN Board. The pathfinder will focus on demonstrating successful resolution of the process, service level and cultural issues that would result from universities sharing data centre space and services. The pathfinder will also seek to demonstrate that potential issues of hosting physical equipment from another university can be avoided via the provision of virtual servers. These potential issues may include: •

Whose insurance covers infrastructure in a shared data centre; the provider university or the host university?

•

What health and safety implications result from hosting another institution’s equipment if a fault occurred and a member of the host institution’s staff were injured?

•

What are the implications around PAT testing?

•

How can standardised remote monitoring be implemented?

•

What are the potential access issues if an emergency site visit is required?

•

What are the potential access issues for delivery and installation of new equipment?

•

How are issues associated with the new Carbon Tax resolved?

From a learning perspective, early agreement will be required on what areas the TPoC and Pathfinder will address. Items that could be considered include: •

How feasible is disk mirroring across the network? Over what distance?

•

How much data can realistically be moved across the network?

•

What service levels are realistic and are these acceptable?

The first stage of the Pathfinder will need to address these issues and provide an agreed way forward on what needs to be measured and how. An additional aspect of the Pathfinder to be considered is the potential for research it could generate. With universities collaborating on providing technical infrastructure, there will be necessary behavioural, cultural and process changes.

12

YHMAN Shared Virtual Data Centre

Business Case Final

At least one Joint Venture university is interested in seeking research funding to support research into certain aspects of the proposed SVDC and the impact it could have on the Research and HE community. The TPoC is under way already, and is set to continue for a number of months. The Pathfinder is expected to last six months. from project inception, which is likely to be within a few weeks of funding being agreed. Issues around summer holidays, clearing and the first few weeks of the autumn term need to be ironed out within this timeframe. A further issue that needs to be identified is the approach to take at the end of the Pathfinder. Stopping all services could have a deleterious effect on the project as a whole, and risk delaying benefit achievement. However, universities need to have the option of withdrawing from the Pathfinder if they find it is not working well for them. There should be some form of process for universities to indicate their willingness to proceed to a live service after three or four months, as the number of likely participants in any future service will have an impact on the likely benefits case of the service. Until the number of likely participants is better understood, and a detailed benefits case is built the likely impacts of a university’s withdrawal from the project after the Pathfinder cannot be fully understood. However, if more than one university were to pull out, then the momentum to move forward into a fully live service would undoubtedly be damaged, along with the benefits case. This scenario would also indicate underlying problems with the service that would need to be identified before implementing it fully.

2.7.2

Region Wide Shared Service The region wide shared service will offer greater access to data centre capacity for those universities that need it; for business continuity, growth, improved data centre facilities, or a combination of the three. This service will be more formal in structure, with contracts and formal service levels in place for YHMAN’s clients. YHMAN will lease spare capacity from those universities that have it, and lease it on to those universities that require it. There are several options to be considered in terms of whose space is utilised and how: •

Several universities have potentially large amounts of data centre space coming on-stream over the next 18 months. In some instances this space is likely to be under-utilised for a period of time, as universities grow into it. This space could be leased by YHMAN to provide it with some or all of the racks it requires in order to be able to provide the SVDC service across the region.

•

One or two universities may have access to space or land that YHMAN could lease at competitive rates and on which it could build its own data centre. Providing a combination of leased and YHMAN owned DC space is the preferred model for provision over the medium to long term, as demonstrated in section 4 below.

13

YHMAN Shared Virtual Data Centre

Business Case Final

As discussed above the regional shared service will offer data centre provision backed up by SLAs and central fault reporting / support. YHMAN could also build its own data centre that would allow several universities to take up space from a central facility, taking advantage of economies of scale and the potential to close down one of their existing data centres. Institutions requiring data centre space will contract directly with YHMAN for access to a certain number of racks rated at a certain power level in a data centre of a minimum standard, which YHMAN will provide for a set price. YHMAN will have agreements in place with all provider universities that cover the staffing, access and lease issues and allow it to offer an all in one service to its customers. This will allow them to see a transparent price per rack which includes a certain level of service. Our analysis below demonstrates the different prices per rack that can be achieved with different combinations of leased space and YHMAN’s own space.

2.7.3

Joint Planning This is not a service offered by YHMAN, rather a collective effort by all the Joint Venture universities to collectively plan their capacity requirements over the forthcoming five to ten years. This will allow a more strategic view to be taken on the overall data centre requirement for the region, and more strategic, targeted, collective investment to be made that reduces the overall cost to the institutions and provides better facilities with a lower carbon footprint.

2.8

SVDC Market Differentiator

2.8.1

Shared Data Centre with a Collective Focus The SVDC will not be a traditional provider of co-location services that anybody can rent rackspace from. It will be a service provided solely with the education market (and at least initially the Higher Education market) in mind. It will be run through the existing and very successful shared services vehicle in the region, YHMAN in which the Joint Venture universities have an equal shareholding. This ownership by and focus on the collective needs of the Joint Venture universities will provide long term path to more efficient provision of better, more resilient data centre services to the Joint Venture universities.

2.8.2

Extremely Fast Connectivity This is a key differentiator for any data centre provision within the region. YHMAN’s network is extremely scalable and resilient due to its topology and the amount of dark fibre in the ground. The amount of bandwidth available across the region via YHMAN is extremely large, and allows the SVDC provision to provide effective service at attractive prices. The cost of commercial bandwidth in commercial co-location centres is an order of magnitude more expensive than YHMAN connectivity, and does not connect onto the JANET backbone.

14

YHMAN Shared Virtual Data Centre

Business Case Final

YHMAN can also offer extension of an organisation’s LAN over the existing YHMAN network into a remote data centre. To offer similar functionality and scale to a university looking to host infrastructure in a commercial co-location provider would be extremely expensive.

2.9

University Data Centre Regional Analysis

Tier‐2  

Data Centre 1  University  University  University  University  University  (DC1)  1   2  3  4  5  Secure  Y   Y   Y   Y   Y   dedicated  space   Water  Y   Y   Y   Y   Y   protection   Air‐con  Y   Y   Y   Y   Y   protection   Structured  Y   N   Y   Y   Y   cabling   Fire  Y   Y   Y   N   Y   protection   Room UPS   Y   N   N  Y   Y   Generator   Status (BC = the  only way to  provide to the  SVDC is by  swapping kit for  BC purposes      Notes: 

Y  

Y  

Y  

Y  

Y  

Y  

Y  

Y  

Y  

N  

Y  

N  

Y  

Y  

N  

N  

N  

N 

Y  

N  

Y  

N  

N  

BC only 

BC only 

8 Racks  

30 Racks 

BC only 

BC only 

BC only 

BC only 

   

   

   

   

   

 

 

 

 

 

   

   

University  University  University  6  7  8  Y   Y   Y  

 

 

 

Need Rack  UPS 

Needs Rack  UPS 

 

Has Diverse  Power  Feeds 

   

Tier‐2  

                  Data Centre 2  University  University  University  University  University  University  University  University  (DC2)  1   2  3  4  5  6  7  8  Secure  Y   Y   Y   Y   Y   dedicated  space   Water  Y   Y   Y   Y   Y   protection   Air‐con  Y   Y   Y   Y   N   New Campus  2009‐14  New Campus  protection   Development  Leased  Development  due 2010/11  Development  Structured  Y   Y   N   Y   Y   cabling   Fire  Y   N   Y   Y   N   protection   Room UPS   Y   Y   N   Y   Y   Generator  

Y  

N  

N  

Y  

N  

15

YHMAN Shared Virtual Data Centre

Business Case Final

Status 

BC only 

BC only 

17 Racks 

9 Racks 

? 

BC only 

7 Racks 

The current position for the Joint Venture universities with respect to their existing data centre provision needs to be looked at from the point of view of both business continuity requirements and future capacity requirements. A further aspect to consider is the carbon footprint of the data centre provision, and how this can be reduced in the coming years, as more attention is paid to it. The table above demonstrates the current situation across the region. It shows that most universities are in need of extra capacity or better capacity or both. Further research demonstrates that the maximum distance between primary and secondary data centres is 2 miles, severely reducing the ability of a secondary site to provide effective BCDR potentially rendering the primary site inoperable. The timeline of potential investment by the universities illustrated below demonstrates that there is likely to be significant investment in DC facilities by the Joint Venture universities over the next three years.

Some of this investment has already been committed; however a significant proportion has not. This demonstrates the need for joint strategic planning to be undertaken, allowing better collective use to be made of data centre investment for the benefit of the whole region.

2.10

Market Analysis

2.10.1

Demand from Universities Initial discussions with IT directors around the region have revealed a desire for more resilient BCDR services and access to more capacity. Some universities, have a pressing need for new DC capacity within the next three years,. Others such as York are pressing ahead with the implementation of brand new DC facilities that will come on stream in the near future. The majority of IT directors feel that they would like to take advantage of shared facilities in the near future, depending on the ability of the YHMAN members collectively, via the SVDC pathfinder, to overcome the process and ways of working issues that have been described above.

16

? 

YHMAN Shared Virtual Data Centre

Business Case Final

2.10.2

Requirements from Universities The key factor which stands out from speaking to the Joint Venture institutions is the internal focus they have on their own requirements when planning. This is entirely understandable, but leads to overprovision of capacity in the short term to allow them to meet long term needs. This leads to inefficient use of DC space for the majority of a DCs life. There is also a direct requirement to improve BCDR facilities and the standards of their existing data centre. The key requirements of the SVDC for individual universities are:

2.11

•

To reduce the overall cost of providing DC capacity

•

Improved BCDR facilities

•

Ensure that the DC capacity offered by YHMAN meets or exceeds the quality standards that a currently enjoyed

•

Ensure more efficient use of existing additional DC capacity.

Legal Constraints The topic of state aid has been raised as a potential barrier to progressing with the SVDC model. KPMG are not in a position to provide legal advice on this topic however the issue needs to be highlighted and described. Any HEFCE funding to support the SVDC business may be considered as state aid. YHMAN operates as an independent company with its own management and board. Its main business will be the provision of services to the HE sector. This will be on a fee paying basis. The data centre market in which YHMAN will operate the SVDC is open to other commercial competition. As such YHMAN could potentially be seen as a company that is funded by a government body and competing unfairly on the open market. KPMG strongly recommend that if legal advice has not already been sought by HEFCE and/or YHMAN on the legality of this approach then it is sought before full implementation is begun.

17

YHMAN Shared Virtual Data Centre

Business Case Final

3

Strategic Fit This section of the business case considers the basis for funding the SVDC and how it would contribute towards achievement of HEFCE’s and the Joint Venture universities’ strategic goals, and the specific benefits the SVDC could deliver to these two groups. It goes on to consider the context within which the pilot phase would operate and potential risks that this may pose to the business. Finally, it considers how the SVDC business might be managed and governed, in order to help achieve buy-in from senior stakeholders.

3.1

HEFCE Goals

3.1.1

Mission HEFCE’s stated mission is: “Working in partnership, we promote and fund high-quality, cost-effective teaching and research, meeting the diverse needs of students, the economy and society.”

3.1.2

Strategic Aims The HEFCE mission statement translates to four core strategic aims: 1. Enhancing excellence in learning and teaching. To ensure that all higher education students benefit from a high quality learning experience fully meeting their needs and the needs of the economy and society. 2. Widening participation and fair access. To promote and provide the opportunity of successful participation in higher education to everyone who can benefit from it. 3. Enhancing excellence in research. To develop and sustain a dynamic and internationally competitive research sector that makes a major contribution to economic prosperity and national wellbeing and to the expansion and dissemination of knowledge. 4. Enhancing the contribution of HE to the economy and society. To increase the impact of the higher education knowledge base to enhance economic development and the strength and vitality of society.

Underpinning these are two cross-cutting supporting aims: 1. Sustaining a high quality higher education sector. To sustain a high quality higher education sector which adapts to the developing needs of stakeholders, and which continues to be recognised as world class. 2. Enabling excellence. To ensure we can effectively deliver our strategic plan, working to the highest standards in all that we do.

18

YHMAN Shared Virtual Data Centre

Business Case Final

3.2

Alignment of SVDC to HEFCE The strategic aims of the SVDC as outlined in section 2.4 are consistent with the strategic aims of HEFCE. The SVDC aims to provide more efficient access to technical resources for the HE community in the region. It promotes collaborative working and strategic planning, to drive out cost efficiencies and improve the provision of DC services. It also promotes improved BCDR and higher availability of services to students, increasing the ability of universities to offer a wider range of services to a wider audience. Longer term, it is envisaged that the SVDC project will implement YHMAN provided data centres that will be built to the latest standards of environmental efficiency. This will dramatically reduce the amount of energy consumed by HE data centres in the region, contributing to HEFCE’s goal of lower energy consumption. This brought about the concept of the Shared Virtual Data Centre (SVDC). Furthermore the SVDC closely parallels the work being investigated by HEFCE into the Shared High End Data Centre (SHED) project that is looking at the potential for more universities to share fewer data centre facilities across the country.

3.3

Replicability A key element of this project from a HEFCE perspective will be how replicable it is across the country, and specifically across other MANs. From a YHMAN perspective, the key factors that make this project realistically achievable are: •

It has been incorporated as a company for a significant amount of time, giving it a track record of trading and providing good levels of service to its customers

•

It has a network that is largely based on its own dark fibre in the ground. This means that the cost of increasing bandwidth between sites or across the network as a whole is generally orders of magnitude lower than would be the case if the network were based on leased lines.

Following some high level analysis of other MANs, there is the potential for the proposed model for YHMAN to be replicated across four other MANs. These four MANs are both incorporated as companies (one of them is run via JANET) and have mainly dark fibre backbones to their network. Two other MANs are already limited companies and have access to high capacity wavelengths via a managed service. Depending on the contracts in place for this managed service, it could be feasible to replicate the YHMAN model in these MANs fairly easily. There are a further five MANs who have dark fibre based networks but who operate as consortia rather than limited companies. This legal status need could be an impediment to successfully implementing the service, but it depends on local situations and the tax and legal advice concerning each situation.

19

YHMAN Shared Virtual Data Centre

Business Case Final

The main factor governing the feasibility of the YHMAN model is not the legal status of the organisation proposing to be the shared service provider; it is the amount of dark fibre in the ground in that particular MAN. The higher the amount of dark fibre, the easier it is to increase the bandwidth between sites to support services such as those being talked about for the SVDC, at relatively marginal cost. It appears as if there is scope for replicability across other MANs around the country, and further investigation of this aspect of the YHMAN SVDC service will be an aspect ofa further report being completed and issued separately later in 2009.

3.4

Alignment of SVDC to the Joint Venture Universities The SVDC project is well aligned to the needs of the Joint Venture universities. It provides improved efficiency, greater collaboration and better DC services. It also presents the opportunity to reduce the carbon footprint of the Joint Venture universities, since more efficient use of data centre space, and new data centre space built to higher environmental standards will result in a reduction in CO2 emissions. There is currently a growing need for data centre space among the Joint Venture universities. At least two universities are at advanced stages of planning new investment in data centre capacity via new-build facilities or refurbishment of existing university space to convert it into a data centre. These facilities are scoped to provide those universities with enough space to expand their capacity over the next few years, meaning the corresponding DCs will not be fully utilised for a large portion of their lives. This represents a cost to those universities. YHMAN can lease the spare capacity from them at attractive rates that at least partially offsets the sunk cost of having a new data centre partially empty. Leasing this space to other universities to allow them to move BCDR facilities off-site or to meet capacity issues of their own would provide efficiencies across the region and be the first step to a fully distributed, virtual data centre offering from YHMAN. Other universities will also require capital investment, of some sort, in new data centre space, over the next five to ten years. Our analysis shows that if these universities collaborate, pool their funding via YHMAN and allow YHMAN to construct a large shared data centre, they will see significant benefits over a five year period. If the universities move towards the implementation of a virtualised environment, these savings are increased further. Joint strategic planning of capacity needs and investment will allow the Joint Venture universities to meet their requirements via YHMAN in the most effective and efficient manner, providing ongoing savings, reducing investment peaks, and laying the foundation for further collaborative working and the provision of further shared services.

3.5

Benefits Financial Benefits The financial benefits for each of the SDVC delivery options are set out in more detail ins section 4 below. In summary, when looking at a scenario where there are five universities requiring a total of 200 racks worth of space, the collective benefits of building a YHMAN data centre to

20

YHMAN Shared Virtual Data Centre

Business Case Final

provide 140 of those racks and leasing 30 racks each from two universities are over £4m over a five year period. Split across the seven universities (five customers of SVDC and two providers to it, the benefits look like this: Distribution of benfits to universities - Scenario 1 Option 3c

Leasing Uni 2 - 30 Racks, £600,000

Uni 1 - 80 Racks, £842,827.69

Leasing Uni 1 - 30 Racks, £600,000

Uni 2 - 60 Racks, £707,120.77 Uni 5 - 10 Racks, £367,853.46 Uni 4 - 20 Racks, £435,706.92

Uni 3 - 30 Racks, £503,560.39

Thus, the more racks a university procures from SVDC in the context of multiple university customers, the greater the benefit to them., with provider universities gaining benefits from such provision and contributing to the overall benefits. If virtualisation is undertaken by the individual universities (i.e. they virtualise their services on their own hardware, hosted in the SVDC), that benefit could rise to a collective £5.5m over the same five year period. Furthermore if virtualisation is undertaken by YHMAN centrally, and YHMAN offers computing capacity to the universities as a service, the corresponding benefits could increase to £10.7m, with individual university benefits rising in line with this. Non financial Benefits The key non financial benefit of implementing the SVDC project is improved BCDR across the region and for individual universities. With secondary infrastructure removed a greater distance from the primary infrastructure it is significantly less likely that a disaster affecting the primary infrastructure will result in total loss of service. The YHMAN network can handle the increased data flows necessary to make this separation work, and can even extend a university’s local network into the remote data centre. This makes the physical distance irrelevant in every case except the rare occasions when an operator needs to physically work on equipment. Another non financial benefit that will flow from the SVDC is the improved coordination between universities on matters relating to commodity items. The joint strategic planning envisaged as part of the project will drive synergies in process execution and sharing of

21

YHMAN Shared Virtual Data Centre

Business Case Final

information in non competitive areas. This should in turn drive greater efficiencies and economies of scale that will feed back as financial benefits in the longer term.

3.6

Strategic Benefits The SVDC proposal provides clear non-financial benefits to three distinct stakeholder groups these are as follows:

3.6.1

Benefits to HEFCE and the UK Government HEFCE and the UK Government will benefit directly from the learning developed in the pathfinder phase of the project on how higher education institutions can collaborate effectively to deliver joint provision of commodity services. The longer term aims of the project will also result in direct benefits to both organisations as the costs to the Joint Venture universities of providing data centre services will fall. This project will also provide an important template for other groups of universities potentially to follow, providing a replicable business model and process for others to use.

3.6.2

Benefits to the YHMAN Joint Venture Universities

3.6.3

Benefits to the Environment Rationalising the number of data centres, and virtualising a significant amount of the current physical infrastructure would dramatically reduce the carbon footprint of the Joint Venture universities. Virtualisation allows more servers to be run on fewer pieces of hardware, reducing the power requirements of the overall infrastructure and the number of physical servers that need to be bought.

3.7

Strategic Success Factors The strategic success factors for this project are as follows: •

Governance and sponsorship - The YHMAN Board and SVDC project board will maintain ownership of the process and business solution throughout the Company life. YHMAN already has effective governance as demonstrated by a significant period of operation. The commitment to the SVDC project can be monitored in this context.

•

The skills and experience developed by the business - This will be measured through the maintenance of documentation developed during project work and the ability to overcome the people and process issues described above.

•

HEFCE support - HEFCE will need to continue to be informed of decisions on all key issues throughout the course of the pilot period and beyond.

•

External stakeholders - YHMAN will manage the expectations of external stakeholders, to ensure that stakeholders “buy-in” to the development of the SVDC. This will be measured through the level of interest in the pilot project from the wider group of Joint Venture universities, and the desire to participate in the live rollout phase of the project.

•

Effective risks and benefits strategies - The project team will implement clear agreed strategies, for the management of key risks and the realisation of benefits if the project is to complete on time and to budget, and meet its objectives

22

YHMAN Shared Virtual Data Centre

Business Case Final

3.8

•

Affordability - This SVDC project will be subject to the budgetary constraints, and will require active management throughout to minimise costs.

•

Project plan - The success of the business will also depend upon the effective management of the execution of the Business Plan, which will enable the Team to reprioritise, anticipate issues and re-engineer the services roll-out timetables as required. Performance of the project plan will be measured against the baseline project plan, resource estimate and key milestone schedules as at the Effective Date.

Strategic Risks and Mitigations Risk

3.9

Mitigation

Ability to manage the staffing structure between host universities and YHMAN

Invest time in talking to staff and seeking their buy-in to doing work in their data centre on another university’s equipment in racks leased by YHMAN.

Shared Service failure

HEFCE should provide as much guidance and support without hindering the organisations. This will be a test of whether shared service can operate successfully in the HEI sector.

Reputational Risk

Make this a separate part of the YHMAN business, with the option of spinning it out into a separate company at a later date.

YHMAN takes up space that it cannot then lease on

Undertake proper capacity planning and match supply to likely demand

YHMAN cannot meet the regional requirements for capacity

Undertake proper capacity planning, as above

Governance and Stakeholders The SVDC business must be cost effective and reliable if Joint Venture universities are going to be convinced that it is preferable to building their own data centre that they control. This is recognised by the project board, who acknowledge the need to provide reassurance to the YHMAN Board and Joint Venture universities that the provision via the SVDC is at least as reliable as their current provision and that they are not exposed to unacceptable levels of risk. This has been considered in relation to deciding on the right corporate vehicle to carry the SVDC project, along with other commercial factors which are described in the following section of this report. In all dealings of the SVDC, the views of the shareholders (namely the 8 Joint Venture universities, led by their respective Vice Chancellors) must be taken into account in determining the appropriate course of action

23

YHMAN Shared Virtual Data Centre

Business Case Final

4

Delivery Options Analysis There are a number of available delivery options for the SVDC project. Initial discussions with YHMAN and HEFCE allowed for the elimination of four of these delivery options as set out in the table below:

Rejected Option Do Nothing

Rationale for Elimination Current issues remain i.e. universities running out of DC space individually, and funding new space independently. New space brought on-line will stand partially unused in its early years (only approaching full utilisation and efficiency towards the end of its life as it becomes full and the cycle begins again). No improvement to current BCDR arrangements unless universities invest in specific BCDR equipment housed in appropriate locations with the corresponding connectivity.

Outsource the entire SDVC requirement to 3rd party data centre providers

This would have the advantage of reducing capital outlay to a fraction of the cost of building new data centre facilities, however the revenue costs of private sector providers will be significantly higher than self-built facilities (in some cases over twice the cost per rack per year). Third party facilities are generally well served by connectivity, but at commercial rates from commercial providers, there would be no YHMAN connectivity into the third party data centre, this would be expensive to install. Without fast, un-contended access to the HE backbone in the region, universities will not be able to utilise their infrastructures effectively, and the costs of providing this access would be high. With the need for extra point to point connections to support the Fibre Channel storage across the network the connectivity requirements into a third party data centre could be increased even further.

Act as ‘clearing house’ for spare data centre space

YHMAN would not take any part in supporting the data centres or the Joint Venture universities, and all agreements would be completely bilateral between the collaborating joint Venture universities. This would not be beneficial to the Joint Venture universities, since the overhead of agreeing contracts and SLAs, in each instance that two Universities chose to share DC space, would be prohibitive. The option would not meet the long term strategic aims of the SVDC project, furthermore the Joint Venture universities have indicated that they would not wish to undertake this approach, and they did not wish this option to be extensively considered in the business case.

24

YHMAN Shared Virtual Data Centre

Business Case Final

Upgrade existing data centre space to form core SVDC provision

Four data centres with the potential for rationalisation and expansion are upgraded to create four facilities of equivalent standard with sufficient space to roll-out the SVDC concept out over a number of years allowing for modern data centre standards to be applied across the selected facilities. There is uncertainty regarding the current state of the data centres and the time and cost that would be required to lift them up to the agreed standard, introducing the risk of delay and overspend. An example of the potential problems is provided by the project to install a new generator and UPS in one of the Joint Venture universities, the project ended up costing over three times as much as initially anticipated (over £1m) due to the need to build a new building to house the generator and UPS. The time, cost and resource constraints imposed by the due diligence that would be required to validate this option properly, would be an unacceptable cost to the project and could mitigate against the overall viability of the SVDC project, making a summer 2009 pathfinder more difficult to achieve since there would be no certainty around which data centres could or would be used as part of the final project.

This analysis below considers the indicative costs and financial benefits of the remaining delivery options for the SVDC. Please note that this analysis has been designed to provide a comparison of the different delivery options, the analysis has not been designed to provide detailed costing against a specific requirement. The comparison of indicative costs and financial benefits for a number of the remaining delivery options for the SVDC has been achieved via three steps:

4.1 4.1.1

•

Step One - Identify high level SVDC requirement scenarios and delivery options, (that combine the different ways of sourcing data centre space)

•

Step Two - Identify indicative costs of sourcing data centre space in different ways (i.e. universities build data centre space independently, YHMAN lease additional capacity from university data centres, YHMAN source data centre space from 3rd party suppliers, and YHMAN build data centre space independently)

•

Step Three - Analyse the indicative costs and benefits of meeting the SVDC requirement scenarios via the different delivery options.

Step One – Requirement Scenarios and Delivery Options Requirement Scenarios To consider the indicative costs and benefits of SVDC project, the following requirement scenarios have been indentified, (please note that for the purposes of this analysis it has been assumed that individual university requirements allow for business continuity and disaster recovery rack space):

25

YHMAN Shared Virtual Data Centre

Business Case Final

4.1.2

•

Scenario 1– Five universities have a combined requirement for 200 new racks with the requirement split: 80, 60, 30, 20 and 10

•

Scenario 2 – Four universities have a combined requirement for 160 racks with the requirement split: 80, 50, 20 and 10

•

Scenario 3 – Three universities have a combined requirement for 120 racks with the requirement split: 70, 30 and 20

•

Scenario 4 – One university has a requirement for 80 racks.

Delivery Options To consider the indicative costs and benefits of SVDC project the following delivery options have been indentified (covering a five year period): •

Option 1 (as is) – The requirements are met entirely by the universities building individual new data centres (please note that Option 1 has been included in the analysis to provide a benchmark against which to measure the financial benefits of the other delivery options).

•

Option 2 – The requirements are met by YHMAN through a combination of racks leased from universities with spare capacity, racks sourced commercially from a 3rd party, and new data centre capacity built by YHMAN. For the purposes of this analysis the combination of leased, sourced and built capacity is split: 25%, 25% and 50%, this would mean that the overall provision is across three sites allowing for adequate business continuity and disaster recovery

•

Option 3 – The requirements are met by YHMAN through a combination of racks leased from universities with spare capacity, and new data centre capacity built by YHMAN. In order analyse Option Three fully, three sub-options have been identified which consider different splits of leased capacity to built capacity, the sub-options being: o

Option 3a – Split: 50% and 50%

o

Option 3b – Split: 40% and 60%

o

Option 3c – Split: 30% and 70% (to the nearest 10 racks).

Option 3a has been selected as the upper limit of leased capacity since it is known that availability of additional space in existing university data centres is limited. Option 3c has been selected as the lower limit of leased capacity since at least 30% of the overall SVDC requirement will be used for business continuity and disaster recovery, and this must be located in a separate location to the main YHMAN data centre. •

Option 4 – The requirements are met by YHMAN through new data centre capacity built by YHMAN in two new data centres, to ensure that adequate BCDR exists.

26

YHMAN Shared Virtual Data Centre

Business Case Final

The four delivery options, (as they relate to each of the requirement scenarios), are described in detail in the tables below:

4.2

Step Two – Indicative Costs of Sourcing DC space

4.2.1

University own provision Traditionally, universities will build a data centre with future capacity and growth requirements in mind. This will lead to a cost that is higher than the costs worked out below, because the fixed costs will generally be higher due to the need to over specify the requirements to allow for growth, providing a high fixed cost then a lower cost per rack as racks are simply added in. Alternatively, universities could find space within buildings on their campuses and refurbish those to required data centre standards. This approach could be cheaper than a new build, although that is by no means certain depending on the access to the site, power, generator and cooling requirements, etc. In the model below, we envisage a modular data centre that will be built to meet the requirements currently, with further modules added as required. This provides lower fixed cost, and a higher cost per rack. The indicative costs for a university to build data centre space independently are as set out in the table in Appendix I below. These costs are based on a number of key assumptions which can be summarised as follows: •

Set up costs are comprised of a fixed cost of £300,000 and a further variable cost of £13,000 per rack, for the purposes of comparison, set-up costs have been amortised over a five year period. This has been worked out from a university who is spending £1.2m on a 70 rack data centre in the near future. Of this £1.2m, around £300,000 is a fixed cost that you could spend on a 10 rack data centre or a 100 rack data centre. This leaves £900,000 as a cost of providing 70 racks. That works out at £13,000 per rack.

27

YHMAN Shared Virtual Data Centre

Business Case Final

4.2.2

•

Total running costs are £4000 per rack per year (previously agreed as a reasonable assumption by the YHMAN Board) which are subject to indexation set at 3% per year

•

Emerging requirements for new University data centres will not exceed 100 racks for individual universities.

YHMAN lease from universities The indicative costs for YHMAN to lease space from universities, (that have additional capacity), are set out in the table in Appendix I. These costs are based on a number of key assumptions which can be summarised as follows:

4.2.3

•

Staffing will cover the effort required to establish and implement leasing arrangements

•

The staffing requirement for YHMAN to manage the leased data centre space will increase non linearly with the amount of space leased

•

The cost to YHMAN to lease data centre space from a University is £4500 per rack per year

•

All running costs are subject to indexation set at 3% per year

•

Available space for YHMAN to lease from existing Universities is limited to 100 racks.

YHMAN source commercially The indicative costs for YHMAN to source data centre space from a 3rd party service provider are as set out in the table in Appendix I below. These costs are based on a number of key assumptions which can be summarised as follows: •

The cost of setting up the service is fixed at £25,000, which has been amortised over a five year period

•

The staffing requirement for YHMAN to manage the service provider are fixed

•

The cost of connectivity is £15,000 per year

•

The ongoing cost of leasing data centre space from the service provider is £9,000 per rack per year

•

All running costs are subject to indexation set at 3% per year.

28

YHMAN Shared Virtual Data Centre

Business Case Final

4.2.4

YHMAN own provision The indicative costs for YHMAN to build data centre space independently are as set out in the table in Appendix I below. These costs are based on a number of key assumptions which can be summarised as follows:

4.2.5

•

Set up costs are comprised of a fixed cost of £300,000 and a further variable cost of £13,000 per rack, as per the costs for a university building its own data centre, which has been amortised over a five year period

•

The staffing requirement for YHMAN to manage the data centre space will increase non linearly with the size of the data centre

•

Running costs are comprised of staffing and rent (for physical floor space), and are subject to indexation set at 3% per year.

Comparison Comparing the different sorts of provision on a cost per rack basis worked out over a five year period provides the following analysis:

YHMAN own provision Total cost per rack per year Size of DC (university self (racks) provided) 10.0 £12,847 20.0 £9,847 30.0 £8,847 40.0 £8,347 50.0 £8,047 60.0 £7,847 70.0 £7,704 80.0 £7,597 90.0 £7,514 100.0 £7,447 110.0 £7,393 120.0 £7,347 130.0 £7,309 140.0 £7,276 150.0 £7,247 160.0 £7,222 170.0 £7,200 180.0 £7,181 190.0 £7,163 200.0 £7,147

Total cost per rack per year (YHMAN lease all from universities) £7,964 £6,690 £7,114 £7,645 £7,199 £6,902 £6,690 £6,530 £6,406 £6,307 £6,168 £6,052 £5,954 £5,870 £5,798 £5,734 £5,678 £5,628 £5,583 £5,543

Total cost per rack per year (commercially provided) £14,198 £11,877 £11,103 £10,717 £10,485 £10,330 £10,219 £10,137 £10,072 £10,021 £9,978 £9,943 £9,913 £9,888 £9,866 £9,847 £9,829 £9,814 £9,801 £9,789

Total cost per rack per year (YHMAN provided from YHMAN DC) £22,744 £12,735 £9,399 £9,006 £7,750 £6,913 £7,043 £6,504 £6,084 £5,748 £5,474 £5,245 £5,051 £4,885 £4,741 £4,616 £4,504 £4,406 £4,317 £4,238

29

YHMAN Shared Virtual Data Centre

Business Case Final

4.3

Step Three – Indicative costs and benefits Four indicators have been used to compare the indicative costs and benefits of each delivery option and each SVDC requirement scenario, these are: •

Total Cost – Based on the combination of data centre sourcing used for each delivery option

•

Capital Outlay – The capital required to implement the delivery option

•

Total Financial Benefit – The savings that options two, three and four offer over option one, combined with the benefit to universities that YHMAN leases data centre space from in order to meet the SVDC requirement

•

Blended Cost per Rack – The cost per rack that YHMAN would be able to offer universities to lease data centre space, based on YHMAN providing a combination of its own provision and outside provision it leases (covering YHMAN set up costs and running costs over a 5 year period).

The following tables and charts set out these indicators as they have been calculated for each of the delivery options and requirement scenarios:

30

YHMAN Shared Virtual Data Centre

Business Case Final

4.3.1 Scenario 1 – Five universities have a combined requirement for 200 new racks with the requirement split: 80, 60, 30, 20 and 10

Comparison of costs and benefits - Scenario 1 £9,000,000 £8,000,000 £7,000,000 £6,000,000

Total Costs

£5,000,000

Capital Outlay £4,000,000

Total Financial Benefit

£3,000,000 £2,000,000 £1,000,000 £0 Option 1

Option 2

Option 3a

Option 3b

Option 3c

Option 4

Comparison of blended cost per rack per year - Scenario 1 £9,000 £8,000 £7,000 £6,000 £5,000 £4,000 £3,000 £2,000 £1,000 £0 Option 1

Option 2

Option 3a

Option 3b

Option 3c

Option 4

Option 1 - Universities provide DC space independently Option 2 - Combination of DC space leased from universities, DC space leased from commercial service providers and DC space from an YHMAN owned DC facility Option 3 - Combination of DC space leased from universities and DC space from a YHMAN owned DC facility, with three different sub-options offering different combinations of these situations (see section 4.1.2) Option 4 - DC space from an YHMAN owned DC facility

31

YHMAN Shared Virtual Data Centre

Business Case Final

4.3.2

Scenario 2 - Four universities have a combined requirement for 160 racks with the requirement split: 80, 50, 20 and 10

Comparison of costs and benefits - Scenario 2 £8,000,000 £7,000,000 £6,000,000 £5,000,000

Total Costs

£4,000,000

Capital Outlay Total Financial Benefit

£3,000,000 £2,000,000 £1,000,000 £0 Option 1

Option 2

Option 3a

Option 3b

Option 3c

Option 4

Comparison of blended cost per rack per year - Scenario 2 £9,000 £8,000 £7,000 £6,000 £5,000 £4,000 £3,000 £2,000 £1,000 £0 Option 1

Option 2

Option 3a

Option 3b

Option 3c

Option 4

Option 1 - Universities provide DC space independently Option 2 - Combination of DC space leased from universities with spare capacity, DC space leased from commercial service providers and DC space from an YHMAN owned DC facility Option 3 - Combination of DC space leased from universities with spare capacity and DC space from a YHMAN owned DC facility, with three different sub-options offering different combinations of these situations (see section 4.1.2) Option 4 - DC space from an YHMAN owned DC facility

32

YHMAN Shared Virtual Data Centre

Business Case Final

4.3.3

Scenario 3 - Three universities have a combined requirement for 120 racks with the requirement split: 70, 30 and 20

Comparison of costs and benefits - Scenario 3 £6,000,000 £5,000,000 £4,000,000

Total Costs

Capital Outlay

£3,000,000

Total Financial Benefit £2,000,000 £1,000,000 £0 Option 1

Option 2

Option 3a

Option 3b

Option 3c

Option 4

Comparison of blended cost per rack per year - Scenario 3 £9,000 £8,000 £7,000 £6,000 £5,000 £4,000 £3,000 £2,000 £1,000 £0 Option 1

Option 2

Option 3a

Option 3b

Option 3c

Option 4

Option 1 - Universities provide DC space independently Option 2 - Combination of DC space leased from universities with spare capacity, DC space leased from commercial service providers and DC space from an YHMAN owned DC facility Option 3 - Combination of DC space leased from universities with spare capacity and DC space from a YHMAN owned DC facility, with three different sub-options offering different combinations of these situations (see section 4.1.2) Option 4 - DC space from an YHMAN owned DC facility

33

YHMAN Shared Virtual Data Centre

Business Case Final

4.3.4

Scenario 4 - One university has a requirement for 80 racks

Comparison of costs and benefits - Scenario 4 £4,000,000 £3,500,000 £3,000,000 £2,500,000 £2,000,000

Total Costs

£1,500,000

Capital Outlay Total Financial Benefit

£1,000,000 £500,000 £0 -£500,000

Option 1

Option 2

Option 3a

Option 3b

Option 3c

Option 4

-£1,000,000

Comparison of blended cost per rack per year - Scenario 4 £10,000 £9,000 £8,000 £7,000 £6,000 £5,000 £4,000 £3,000 £2,000 £1,000 £0 Option 1

Option 2

Option 3a

Option 3b

Option 3c

Option 4

Option 1 - Universities provide DC space independently Option 2 - Combination of DC space leased from universities with spare capacity, DC space leased from commercial service providers and DC space from an YHMAN owned DC facility Option 3 - Combination of DC space leased from universities with spare capacity and DC space from a YHMAN owned DC facility, with three different sub-options offering different combinations of these situations (see section 4.1.2) Option 4 - DC space from an YHMAN owned DC facility

34

YHMAN Shared Virtual Data Centre

Business Case Final

4.3.5

Summary of the Indicative Costs and Benefits Option 1 – In each scenario, (with the exception of Scenario 4), the Total Cost and Capital Outlay are highest, this is driven by the set up costs of building a new data centre for each university. Option 2 – In each scenario, (with the exception of Scenario 4), the Total Costs are second highest and Total Financial Benefits (over Option 1) are lowest, this is driven by the prohibitively high costs of sourcing data centre space commercially (typically £9,000 per rack per year). Option 3 – In each scenario, Option 3 offers the most compelling combination of Total Costs and Total Financial Benefits. Of the sub-options 3a to 3c, 3a delivers the highest benefits and 3c offers the lowest costs. For 3a this is driven by the greater additional benefit to universities with additional capacity that lease space to YHMAN, for 3c this is driven by the lower running costs of YHMAN owned data centre space versus university owned data centre space. The magnitude of the benefits offered by Option 3 also increases as the number of universities and racks comprising the overall SVDC requirement increases. Option 4 – In each scenario, Option 4 offers costs and benefits that sit between Option 2 and 3. This is driven by the requirement for YHMAN data centres to be built in two locations in order to provide adequate BCDR (since the data centre space would be sourced from YHMAN alone).

4.3.6

Distribution of benefits to universities The benefits identified above would be distributed to universities by YHMAN offering a lower cost per rack than the universities are able to achieve by building and running data centres independently, and (in the case of Option 2 and Option 3), by YHMAN leasing data centre space from universities with additional capacity. By example the chart below illustrates the distribution of benefits to universities for Scenario 1, Option 3c: Distribution of benfits to universities - Scenario 1 Option 3c

Leasing Uni 2 - 30 Racks, £600,000

Uni 1 - 80 Racks, £842,827.69

Leasing Uni 1 - 30 Racks, £600,000

Uni 2 - 60 Racks, £707,120.77 Uni 5 - 10 Racks, £367,853.46 Uni 4 - 20 Racks, £435,706.92

Uni 3 - 30 Racks, £503,560.39

35

YHMAN Shared Virtual Data Centre

Business Case Final

Clearly in this scenario university 1 realises the most benefit from sourcing its requirements from the SVDC, with the other universities realising decreased benefits as their needs decrease. Those universities who lease racks to YHMAN for SVDC provision will also realise benefits proportional to the number of racks they lease to YHMAN. It should be noted that the distribution of benefits to universities will not be proportional to rack space requirements, since universities with greater requirements would be more cost effective in meeting those requirements independently and therefore will see less benefit (proportionally) in moving to the SVDC/YHMAN model.

4.4

Virtualisation by Universities By virtualising data centre equipment it is possible for universities to reduce their corresponding requirement for data centre rack space by up to 50%. To analyse the impact that university virtualisation would have on the SVDC project a further delivery option has been identified: Option 3d – The SDVC requirements are met by YHMAN through a combination of racks leased from universities with spare capacity, and new data centre capacity built by YHMAN (split 30% leased and 70% built). In each case the data centre equipment would be virtualised over a period of time based on the following assumptions: •

Virtualisation would be achieved over a three year period with a 30% rack space saving from year three and a 50% rack space saving from year four

•

The cost to virtualise data centre equipment would increase the set-up cost for the YHMAN data centre by £100,000

•

By the time that virtualisation is achieved, there is demand for the additional YHMAN capacity that becomes available.

The impact of Option 3d on individual universities would be that fewer racks are required from year three onwards, allowing other universities to utilised data centre rack space that becomes available, as set out in the table below (based on the Scenario 1 requirement):

This results in direct financial benefit to the universities, since fewer racks are required from YHMAN from year three onwards.

36

YHMAN Shared Virtual Data Centre

Business Case Final

The indicative costs and benefits for Option 3d and Scenario 1, (as compared the indicative costs and benefits for Scenario 1, Options 1, 2, 3a, 3b, 3c and 4) are as set out in the table and charts below:

Comparison of costs and benefits - Scenario 1 £9,000,000 £8,000,000 £7,000,000 £6,000,000 Total Costs

£5,000,000

Capital Outlay

£4,000,000

Total Financial Benefit

£3,000,000 £2,000,000 £1,000,000 £0 Option 1

Option 2

Option 3a

Option 3b

Option 3c

Option 3d

Option 4

Comparison of blended cost per rack - Scenario 1 £8,000 £7,000 £6,000 £5,000 £4,000 £3,000 £2,000 £1,000 £0 Option 1

Option 2

Option 3a

Option 3b

Option 3c

Option 3d

Option 4

Option 1 - Universities provide DC space independently Option 2 - Combination of DC space leased from universities with spare capacity, DC space leased from commercial service providers and DC space from an YHMAN owned DC facility

37

YHMAN Shared Virtual Data Centre

Business Case Final

Option 3 - Combination of DC space leased from universities with spare capacity and DC space from a YHMAN owned DC facility, with three different sub-options offering different combinations of these situations (see section 4.1.2 and section 4.4) Option 4 - DC space from an YHMAN owned DC facility

38

YHMAN Shared Virtual Data Centre

Business Case Final

The chart below illustrates the distribution of benefits to universities for Scenario 1, Option 3d: Distribution of Benefits - Scenario 1 Option 3d

Leasing Uni 2 - 30 Racks, £600,000 Uni 1 - 80 Racks, £1,413,812.59 Leasing Uni 1 - 30 Racks, £600,000

Uni 5 - 10 Racks, £439,226.57

Uni 4 - 20 Racks, £578,453.15

Uni 2 - 60 Racks, £1,135,359.44 Uni 3 - 30 Racks, £717,679.72

The analysis of the results indicates that there is a clear trend towards increased savings as more universities join in the SVDC service with this saving being accentuated when they start to move towards virtualisation and provision of a virtual environment, reducing their requirement for physical rack space.

4.5

Virtualisation by YHMAN By purchasing DC hardware and virtualising it centrally, it would be possible for YHMAN to extend the service offering that they provide to universities as part of the SVDC project, and lease universities access to DC hardware to meet their specific computing requirements. To analyse the impact that YHMAN virtualisation and the leasing of DC hardware to universities would have on the SVDC project, a further delivery option has been identified: Option 3e – The SDVC requirements are met by YHMAN through a combination of racks leased from universities with spare capacity, new data centre capacity built by YHMAN (split 30% leased and 70% built), running YHMAN hardware offering virtual services to the Joint Venture universities. In each case the university’s infrastructure would be virtualised prior to the service being offered, based on the following assumptions: •

75% of the total SVDC hardware requirement could be virtualised achieving a compaction ration of 10:1 for CPUs

39

YHMAN Shared Virtual Data Centre

Business Case Final

•

YHMAN would be required to source less total hardware than universities sourcing hardware independently (providing an additional benefit over and above the previously considered options)

•

YHMAN would be required to provide less DC rack space since there would be less hardware to house in the DC

•

The power requirement for virtualised DC space (approx 8 KW per rack) is greater than the power requirement for non-virtualised DC space (approx 3.5 KW per rack) which results in slight reduction in the benefits of YHMAN virtualisation. This is a conservative view, as evidence suggests that virtualisation does reduce overall data centre power consumption.

The indicative costs and benefits for Option 3e and Scenario 1, (as compared the indicative costs and benefits for Scenario 1, Option 1) are as set out in the table and chart below, (please note that these figures have not been compared to the other options under Scenario 1 since they include DC hardware costs, which have not been included under any of the other delivery options):

Comparison of costs and benefits - Scenario 1 £35,000,000 £30,000,000 £25,000,000 Total Costs

£20,000,000

Capital Outlay £15,000,000

Total Financial Benefit

£10,000,000 £5,000,000 £0 Option 1

Option 3e

Option 1 - Universities provide DC space and hardware independently Option 3e - Combination of DC space leased from universities with spare capacity and DC space from an YHMAN owned DC facility with virtualised DC hardware provided by YHMAN

40

YHMAN Shared Virtual Data Centre

Business Case Final

The benefits identified above would be distributed to universities by YHMAN offering a lower cost for DC services than the universities are able to achieve by building and running data centres using their own hardware, and by YHMAN leasing data centre space from universities with additional capacity. The chart below illustrates the distribution of benefits to universities for Scenario 1, Option 3e: Uni Benefits 3e - YHMAN Virtualisation Leasing Uni 2 - 15 Racks, £300,000 Leasing Uni 1 - 15 Racks, £300,000 Uni 5 - 10 Racks, £760,070

Uni 4 - 20 Racks, £1,220,140

Uni 1 - 80 Racks, £3,980,559

Uni 3 - 30 Racks, £1,680,210

Uni 2 - 60 Racks, £3,060,419

4.6

Recommendations

4.6.1

Technical Proof of Concept to investigate potential technical issues The Technical Proof of Concept will investigate the technical issues of running data centre services from remote sites. Sheffield Hallam University and Leeds University have already put the technical elements in place and have begun testing some services across the infrastructure. As this testing progresses, information about the capacity of the service, ability to replicate data over distance and other significant aspects of the service will be generated. This will feed into the Pathfinder requirements and the design considerations for the full live service.

41

YHMAN Shared Virtual Data Centre

Business Case Final

4.6.2

Define sourcing route for Pathfinder Hardware To maximise the success of the Pathfinder, access to appropriate hardware will need to be sourced. The table below highlights the potential costs to YHMAN of procuring access to new equipment to support the Pathfinder. There are several different potential routes to source this hardware for the six month period: •

Lease the hardware from a supplier or reseller for six months

•

Collectively procure the equipment for use within individual Joint Venture universities, with each university loaning some of the equipment to YHMAN for the six month period before taking it back into their organisation

•

Work with vendors to encourage them to loan the requisite equipment for the period of the Pathfinder

•

Re-use existing hardware if none of the above options is seen to be viable.

Agreeing appropriate hardware specifications and the services to be provided on that hardware will need to be agreed before moving into the Pathfinder stage.

4.6.3

Pathfinder to Investigate the Implementation of the SVDC service Our analysis indicates that the universities will benefit significantly from the SVDC service, and that the greater the service required, the greater the benefits for the participating organisations. We therefore recommend that the service be initiated with a small scale pathfinder project, centred on delivery options that consider both university and YHMAN virtualisation, to be implemented in summer 2009. The pathfinder project will provide significant insight into the process, cultural and technical issues thrown up by offering shared services in this area. The proposed pathfinder involves using equipment in host universities to offer a virtual service to other Joint Venture universities. The ‘customer’ universities would receive a secure, discrete service of remote access to a number of virtual servers that they could administer. This will prove that the concept of virtualising infrastructure off-campus is a valid one that offers the potential for savings over the longer term. It will also allow YHMAN to define the processes and structures necessary for offering a service such as this, and understand the barriers that need to be overcome for the service to be a success. The pathfinder will not be focused on providing cost efficiencies or medium to long term financial payback. Rather it will focus on demonstrating successful resolution of the very real and immediate process, service level, and people issues that need to be resolved to move forward with a wider implementation. Environmental issues will be a key focus at this stage. Making data centres more efficient and reducing their carbon impact will allow the Joint Venture universities to address their responsibilities in this area, and provide relevant learning for the rest of the sector. An additional aspect of the Pathfinder to be considered is the potential for research it could generate. With universities collaborating on providing technical infrastructure, there will be necessary behavioural, cultural and process changes.

42

YHMAN Shared Virtual Data Centre

Business Case Final

At least one Joint Venture university is interested in seeking research funding to support research into certain aspects of the proposed SVDC and the impact it could have on the Research and HE community. The pathfinder will cost up to an estimated £215,000, with the setup and evaluation phase anticipated to be around six months. We recommend HEFCE funding is sought to run the pathfinder, which will cover the costs of new equipment needed to run the virtual servers, individual university staff time to configure the new equipment, and YHMAN project staff to provide the YHMAN support to the pathfinder. Also included in this cost is the external support that will be required to set up the pathfinder. This support will include the setting up of the programme management functions including the governance and reporting arrangements, process definition and implementation, service management aspects, based on ITIL v3 and working with HMRC to clarify and confirm the tax regime that the new service will be covered by.

4.6.4 Detailed pathfinder costs Hardware Item Amount required Cost Servers 4 £115,000 VMware ESX licenses 16 £27,600 Storage option 1 (Openfiler) 2 £9,500 Storage option 2 (SANMelody) 2 £46,000 SASBeast disk arrays 2 £115,000 Total capital cost. This does not £269,100 to £303,600 depending on storage choice plus network infrastructure. represent the actual cost of the pathfinder; merely the up front cost of the hardware. Pathfinder costs are below, and are for access to the hardware for the 6 months Pathfinder period. Cost per annum when amortised £53,820 - £60,720 over five years Cost for access to hardware £26,910 - £30,360 for the six months of the Pathfinder These costs are based on the total capital cost of the hardware. The average life expectancy of infrastructure hardware is around five years. Therefore the amortised cost of the hardware on an annual basis is between £54,000 and £60,000. The Pathfinder is expected to last six months, with the hardware required for around four months of that. Therefore the expected hardware costs for the Pathfinder are estimated at between £20,000 - £30,000. One option to be considered is the leasing of this equipment from a hardware provider for the period indicated. This will need to be carefully managed to minimise the threat to any potential hardware procurement further down the road and we would recommend seeking formal procurement advice before engaging with any suppliers at this stage.

43

YHMAN Shared Virtual Data Centre

Business Case Final

Another option would be for a university who would be requiring similar hardware to purchase it in time for the Pathfinder, then lease it to YHMAN for the duration of the Pathfinder for the costs indicated above. We recommend ascertaining the feasibility of the hardware aspect of the Pathfinder early on, as access to appropriate hardware will play an important role in determining the success or otherwise of the Pathfinder. Should access to new hardware be difficult, it should be determined whether older hardware would be appropriate for the Pathfinder stage. At the end of the Pathfinder they would then have the option of selling it on to YHMAN, or taking it into service for their own purposes. Internal Staffing We recommend employing a Project Manager to manage the delivery of the Pathfinder project. This person will be expected to liaise with the external support and drive the project forward in conjunction with them. The anticipated cost for this person for the Pathfinder phase is estimated to be £25,000 (£30,000 per annum, plus on costs, for a six month period). Additionally, we expect to require input from internal university technical staff. It is difficult to estimate the total amount of time required from internal university staff at this stage, but it is unlikely to be lower than one day per week. We therefore recommend budgeting £10,000 per university to recompense them for the staff time they are losing to the project. With four universities involved in the pilot, that would be £40,000

External Support Support area No. Days Cost Pathfinder initiation 2 £3,000 Programme Management 5.5 £9,000 Service Management Design 15 £26,500 Process Design 9.5 £11,500 Total external support cost £50,000 These costs are indicative and subject to change, subject to agreement of a formal programme of work. Exit strategy As the pathfinder builds momentum and the service management and programme structures are put in place, it can be slowly built up into a larger service, based on the costs and benefits outlined below with a defined charging and service provision mechanism. This will allow the universities who are not participating in the pathfinder to begin to take services from YHMAN, and it will allow the pathfinder universities to

44

YHMAN Shared Virtual Data Centre

Business Case Final

increase their participation from the small scale pathfinder involvement to a larger scale commercial service provision. Universities wishing to exit from the Pathfinder when it finishes will need to have a means of doing so. We recommend that those universities who will not wish to carry on with the service should indicate as soon as possible that they will be exiting. This will allow the final business case to be built around a definitive set of figures. Universities should indicate after five months whether they wish to continue with the service, or leave at the end of the Pathfinder, and universities wishing to join should also indicate their desire to do so at this stage. A special Board meeting should convene after this point to examine the Pathfinder and assess how successful it has been, and how viable the service is long-term, based on the numbers of institutions who have indicated they would like to sign up for the live service. That meeting should take the decision as to whether the live service is implemented, and how it should be funded. This decision point will also determine the shape of the service, and whether YHMAN is looking to build its own data centre in the near future or not. Slowly increasing the scale of the service mitigates the risks of implementing a large scale service on day one and then facing significant implementation problems that destroy confidence in the service and set the project back a number of years. Our analysis indicates that a combination of Joint Venture university capacity leased to YHMAN and YHMAN’s own data centre provide the best value for money for universities looking to take up the service, and the largest overall benefits. We therefore recommend that as the service grows in scale, this combination is the medium term aim, to allow the maximum benefits to flow from the project to the Joint Venture universities. It should be noted that there is a parallel HEFCE funded project; the Shared High End Data Centre (SHED) project. That project is similar to this one, and the learning generated by each should be jointly shared on a regular basis, to allow emerging best practice to flow between the two projects and allow them to build on each other’s successes. This will cover areas such as the service management aspects, where Event, Problem and Incident management processes will be largely replicable, as will SLAs potentially.

45

YHMAN Shared Virtual Data Centre

Business Case Final

4.6.5

Joint strategic planning In parallel with the pathfinder project described above, it is recommended that that the Joint Venture universities and YHMAN carry out an extensive joint planning exercise, the objectives of which will be: 1. To understand the medium and long term DC capacity requirements of the individual Joint Venture universities 2. To build detailed cost estimates for YHMAN provided DC space and hardware 3. To develop a detailed service model for the proposed YHMAN services including pricing and SLAs 4. To plan implementation and transition to the live service. These planning activities will necessitate a fully resourced project that builds senior stakeholder engagement, and allows the proposed YHMAN services to be developed in a realistic manner that will be collectively accepted and agreed to by HEFCE, YHMAN and the Joint Venture Universities. This work will also require external support, and the agreement from all the participants that they will provide resource for at least one day per week who will drive the work forward. The external support costs are as follows: Phase

Total days

Total cost

Setup

2.0

£2,500

Programme Management

5.5

£9,000

Workshop facilitation

11.0

£12,500

Benefits identification + management

20.5

£26,500

Tax advice

10.0

£19,500

Totals

49

£70,000

These costs are indicative and subject to change, subject to agreement of a formal programme of work.

46

YHMAN Shared Virtual Data Centre

Business Case Final

4.6.6

Total estimated costs for Pathfinder and Joint Strategic Planning setup work Cost Element

Cost

Pathfinder Project Manager

£25,000

Internal university staff costs

£40,000

Access to Pathfinder Hardware

£20,000 - £30,000

External support for Pathfinder

£50,000

External support for Joint Strategic Planning

£70,000

Total

£205,000 - £215,000

The total costs for the two pieces of setup work for the main SVDC service are estimated to be approximately £215,000. These break down into £120,000 for the combined Joint Strategic Planning and Pathfinder work external support; £25,000 for a Pathfinder project manager; £40,000 for internal university staff costs in supporting the Pathfinder; and a maximum £30,000 for access to hardware for the pathfinder. If the hardware costs were reduced, or legacy hardware were used, the costs could be reduced further. It is likely that if the two exercises are undertaken separately, costs could increase, as some of the work could well be duplicated across the two activities.

47

YHMAN Shared Virtual Data Centre

Business Case Final

5

Commercial Considerations YHMAN is already a successful business, with a long history of providing network services to the eight Joint Venture universities. It has a proven business model of charging for the services it provides and retaining surpluses year on year to provide for future innovations and reducing the risks to the Joint Venture universities. The SVDC model would present different challenges to the networking model. Different universities will want different services at different levels of availability and different types of pricing. For this reason, the options available for Shareholder Universities will be restricted. The range of data centre services available will be restricted to those that are most relevant to the majority of shareholder universities, and will be re-validated annually. The services described in this document have been validated by the SVDC project board and the YHMAN Board as those they would like to see as part of the SVDC. Capitalisation of YHMAN to build a data centre could be achieved in a number of different ways: HEFCE grant; HEFCE loan; loan from the Joint Venture universities; capitalisation from the Joint Venture universities; YHMAN’s own reserves; or a combination of these. The capital cost of constructing the YHMAN DC(s) has been amortised over five years and included in the cost per rack calculations, meaning the capital cost of building a YHMAN data centre is recovered within five years.

5.1

Costs and Affordability Costs and affordability of the SVDC project vary according to how many universities collaborate to provide the service via YHMAN, our analysis shows that for one or two universities the benefits would be marginal, however for five universities, with a combined requirement for 200 racks of DC space, the combined financial benefits could be as much as £4m over a five year period. If virtualisation is undertaken by the individual universities, that benefit could rise to a collective £5.5m over the same five year period. Furthermore if virtualisation is undertaken by YHMAN centrally, and YHMAN offers computing capacity to the universities as a service, the corresponding benefits could increase to £10.7m.

5.2

Consequences of Not Proceeding The failure to implement the Shared Virtual Data Centre project will result in higher costs being borne by the Joint Venture universities. Each university will have to make a large capital investment in data centre facilities or a large ongoing revenue commitment to private sector data centre providers to allow them to keep up with demand for IT services from their user bases. Any purpose built facilities are likely to be partially empty for a large portion of its life, as they will be built with at least a ten year time horizon in mind. This is not only a poor use of HEIs’ capital resources, it does not align with the increased need for environmentally aware policies across the HE sector and beyond.

48

YHMAN Shared Virtual Data Centre

Business Case Final

While there are new techniques and products available in the area of data centre construction such as modular data centres or high density self contained data centres, these approaches would not entirely eliminate the issue of unused space. Alongside the issue of large capital investments is that of business continuity. Even with investment in a new facility to give them greater capacity, the joint Venture universities would still be faced with two unpalatable choices:

5.2.1

•

Restrict investment to the extended capacity and risk large scale outages of service in the event of a natural disaster, or

•

Invest even more money in the relocation of their business continuity facility to a geographically distant site.

Market Segmentation YHMAN will deliver the SVDC initially only to the Joint Venture universities. The pathfinder phase will see four Joint Venture universities partaking in a trial of distribution of business continuity infrastructure via the SVDC. The next phase will see all those Joint Venture universities that wish to use the service leasing rackspace from YHMAN. The project will still be focused on the Joint Venture universities, and it is unlikely to open up beyond them for a number of years, until the proof that service levels and logical separation of data can be maintained is available. Longer term, the service could be opened up to FE colleges across the region, and indeed any UK Public Sector organisation that currently receives its JANET connection via YHMAN.

49

YHMAN Shared Virtual Data Centre

Business Case Final

6

Corporate Structure Options appraisal

6.1

Summary An analysis of possible structures was undertaken based on YHMAN or an associated company providing the service. The company structure was assumed to be a company limited by shares as this conveys certain advantages: •

The use of a limited company affords some protection

•

The company’s intention is to operate on a profitable basis to allow it to be in a position to extend the amount of data centre space available across the region as that becomes necessary

•

A CLS provides full flexibility for the Joint Venture universities in extending the provision or expanding the company in the future by allowing new shareholders to be created with shares that carry a different balance of equity and voting rights.

KPMG are currently working with HEFCE to explore if shared service provision within the public sector can be exempt from VAT. Article 132(f) of the European VAT Directive is currently not enacted in UK legislation but HMRC have confirmed that UK taxpayers can rely on applying the article directly to their services. Due to the ambiguous nature of the wording of article 132(f), and the uncertainty as to HMRC’s interpretation of the article it is considered that the proposed SVDC service offering would not qualify for exemption. If HMRC clarify their interpretation of article 132(f) and/or article 132(f) is enacted in UK legislation and it allows for exemption in certain circumstances to apply to shared service provision within the public sector, there will be an immediate benefit to most public sector clients by removing a VAT charge which is usually a cost, especially to HEI’s. It is recognised that if exemption can not be applied, then this will mean that VAT will need to be charged on any services supplied to Universities. Although VAT is normally irrecoverable by HEIs depending on their special exemption method it is anticipated that this may not be a material additional cost and therefore it would not act as a disincentive for universities to use the services of the company, since the price points have been set to easy purchasing decisions rather than maximising revenues. The facility exists for a gift aid payment to be made to the University partners to shelter any corporation tax due on profits, after repayment of the loan has been made, but the commercial implications of this route will require consideration as any gift aid payment would adversely affect the cash reserves of the company.

6.2

Corporate Structure Options Appraisal Various corporate structures for delivering the SVDC service have been considered. These included: •

A cost sharing association incorporating an entirely new standalone company

•

Forming a new company linked directly to YHMAN

50

YHMAN Shared Virtual Data Centre

Business Case Final

•

Maintaining the existing YHMAN structure and offering the SVDC as an extra service of YHMAN.

These options were considered in the context of a number of different factors that could affect the new service, the organisation delivering it and the Joint Venture universities. These included the:

6.3

•

Ability of the new company to raise capital from its shareholders to facilitate the building of extra data centre space

•

Tax efficiency of delivering the SVDC services to the Joint Venture universities

•

Impact of the new structure on the Joint Venture universities

•

Impact of the new structure on YHMAN

•

Ability of the new structure to foster new partnerships across the region in the future

•

Perception of customers, users and other stakeholders

•

Mitigating of risk

•

Stakeholder agreement

•

Staffing requirements, employment rights and secondment model.

Advantages and Disadvantages of Corporate Structure Models Considered

Corporate Structure A cost sharing association

Advantages 1. Allows the Joint Venture universities to share the costs and overall provision of the SVDC 2. Allows the services provided to be viewed as VAT exempt as they would not be provided on a commercial basis

Disadvantages

1

Requires the services that are being exchanged to be equal in nature, making it difficult for one or two larger universities to provide service to the others

2

Is not an appropriate vehicle for providing a commercial service

51

YHMAN Shared Virtual Data Centre

Business Case Final

Incorporating an entirely new standalone company

Forming a new company, linked directly to YHMAN

Maintaining the existing YHMAN structure

1

Keeps the SVDC provision discrete from the YHMAN networking provision

1

Is a completely new company, with no history of trading, providing services or running a surplus

2

Insulates YHMAN from the risks of undertaking new services

2

Would place an extra burden of responsibility on the Joint Venture universities’ representatives who would have two sets of responsibilities to discharge; to YHMAN and the new company

1

Keeps the SVDC service at arms length from the established YHMAN service

1. Adds extra complexity into the service provision

2

Allows YHMAN to be involved in the service provision

1

Maintains the relationship directly with YHMAN, providing confidence from YHMAN’s record of working with the Joint Venture Universities

2

2. Potentially removes direct involvement in the SVDC from the Joint Venture universities, with YHMAN mediating in the middle 1. Exposes YHMAN to potential risks from any problems with the SVDC provision 2. Potentially distracts YHMA from its core business of network provision

Reduces the difficulty of carrying a surplus from one year to the next, as YHMAN has a history of doing so

The discussions centred on whether it was better to incorporate an entirely new company or offer the SVDC as a new service direct from YHMAN. The final recommendation was to settle for YHMAN offering SVDC as a new service, with no change to corporate structure. There were several reasons for doing this: •

YHMAN has a successful history of working closely with the Joint Venture universities, and asking them to procure data centre services from a new company they hadn’t traded with before was felt likely to be difficult

•

YHMAN has carried a year on year surplus into the next year for a while now, and can point to having done so for differing investment reasons. A new company would find it more difficult to be able to carry that surplus without incurring potentially significant tax on it, and this would make it more difficult to capitalise the new company in order for it to procure a new data centre or lease racks from Joint Venture universities

•

Working with HMRC to decide any tax exemption or otherwise for the SVDC service would be easier via YHMAN, which has been incorporated for a long time than a new company only just incorporated

52

YHMAN Shared Virtual Data Centre

Business Case Final

•

The burden on the joint Venture universities would be increased potentially significantly with the incorporation of a new company; giving them another directorship to administer. Allowing YHMAN to offer SVDC as an extra service would mitigate against this increase in responsibilities and burden

•

Incorporating the service into YHMAN would make the provision of networking included in new data centre services a more straightforward process and would not involve the SVDC company having to buy (and pay VAT on) networking services from YHMAN.

6.4 Implications of VAT on the SVDC The VAT on the SVDC services is a complex area that touches all aspects of the service relationship with the Joint Venture universities as well as each Joint Venture university’s partial exemption method with HMRC. For example, if YHMAN leases a section of the data centre, to include access to (but not actual provision of) power and cooling, and that section of the data centre is designated as YHMAN only this letting of space may be exempt from VAT both to YHMAN and by YHMAN when it charges a user university. If the host institution needs to place equipment in that area, it would have to lease it from YHMAN just like any other institution would. The ability of a host university to exempt the lease or licence to occupy granted to YHMAN will depend upon whether the host university has opted to tax its data centre. If it hasn’t opted then the rent charged to YHMAN for the letting of space could possibly be exempt from VAT, thus potentially allowing YHMAN to exempt part of its charge to a user university. If the host university has opted to tax its data centre its rent charge to YHMAN will be liable to VAT, in which case YHMAN would either also have to opt to tax and charge VAT to the user university or, if it does not opt to tax, treat the VAT charged by the host university as non-recoverable and not charge VAT to the user university. Charges by the host university to YHMAN and by YHMAN to a user university for “other services” i.e. anything in addition to the renting of space, would be liable to VAT. YHMAN will be able to reclaim all the VAT it pays whereas the universities will only be able to reclaim the amounts relevant to their partial exemption method. If upgrades are being carried out to data centres for use by YHMAN, it invariably will be better for YHMAN to contract with the contractors directly. This is because if the host institution contracts with the contractors it may only be able to reclaim the amount of VAT relevant to that institution’s partial exemption method whereas YHMAN would be able to obtain full recovery of the VAT on the contractors’ charges if it opts to tax the data centre. Any lease or licence to occupy granted to YHMAN by a host university should ideally give it the right to alter premises to its specification. If the possibility of having a separate (VAT exempt) charge for the letting of space is to be pursued we recommend HMRC’s agreement to this treatment is obtained before implementing this route. In order to do so it is imperative that a full service definition is produced that outlines all the services provided by YHMAN and the host universities.

53

YHMAN Shared Virtual Data Centre

Business Case Final

Currently discussions are ongoing between the North Eastern Universities Purchasing Consortium (NEUPC) and HMRC around the implications of EU Directive 132f and whether services provided by members, that are purely for the internal benefit of members should be VAT exempt. There has to date been no acceptance of this position from HMRC, but discussions are ongoing, and we recommend that YHMAN keeps abreast of developments as the SVDC project develops, to allow you to maximise the opportunity of benefiting from any potential ruling from HMRC that favours VAT exemption.

6.4.1 YHMAN as provider of Data Centre services In this scenario YHMAN offers Data centre services as an extra service on top of the networking services it offers to the Joint Venture universities and other educational institutions. In terms of VAT and other tax considerations this has several advantages: •

YHMAN has a history of providing paid for services to the regional HE community

•

It has been VAT registered for a number of years

•

As it is already VAT registered it will be easier to work with HMRC to obtain any potential VAT exemption for the space element of the SVDC services

•

It has a history of carrying surpluses over from one year to another for investment in longer term projects

•

The successful history of the joint venture universities and YHMAN provides a solid foundation for continued and broader provision of services to the Joint Venture universities

The only difference in the VAT registration of YHMAN would be that it potentially would lease space (racks) in data centres to universities without VAT and therefore treat as non-recoverable some of the VAT on its general overheads. This option for providing the data centre service would be the preferred option in our opinion.

6.4.2 The Joint Venture universities form YHDATA to provide the SVDC service In this scenario, a new company, YHDATA, would be formed specifically to provide the SVDC service. There are some major disadvantages to the joint Venture universities proceeding down this route: •

YHDATA would have no history of trading services, with no history of surplus funds to invest

•

YHDATA would therefore require funding (either via the member Universities or YHMAN itself) and this would give rise to tax issues which would need to be addressed.

•

YHDATA would have to go through the VAT registration process with HMRC before being able to determine whether its services would have any VAT exemption

•

The flow of service and payments will be more complicated. The universities will buy data centre services from YHDATA and networking services from YHMAN.

54

YHMAN Shared Virtual Data Centre

Business Case Final

•

YHDATA will also need to buy networking services from YHMAN to support the regional data centre service. These charges will also attract VAT, although YHDATA will be able to reclaim this VAT. VAT on these charges could be avoided if YHDATA and YHMAN register for VAT as a VAT group registration. This however would mean YHMAN’s current VAT registration would have to be cancelled and a new group registration formed with a new VAT number.

•

The extra burden on the Joint Venture universities of directing two companies would be potentially significant, and the cost and extra administration of two companies would be higher

•

An additional company would reduce the profit at which higher rate tax would be paid as the upper limit is divided by the number of associated companies.

We do not believe this model is the preferred option for delivering the SVDC due to YHMAN being in a better position to deliver it.

6.4.3 YHMAN becomes a grant giving and receiving organisation In this scenario, YHMAN would gain the ability to receive & award grant payments. It would then: •

Pay service provider universities for DC space & operations etc. support with a grant.

•

Receive grant payments from service client universities

These payments would be grant payments intended to allow the universities or YHMAN to improve their business continuity and IT infrastructure arrangements. We do not recommend YHMAN and the Joint Venture universities adopt this charging model for the SVDC services. We believe that HMRC would view the lease of data centre space from the provider universities by YHMAN and the provision of data centre services to other universities, facilitated by grant awards in either direction, as a series of barter transactions. They would therefore attach a value to these transactions and levy VAT on that value. This would complicate the overall service charging arrangements, and potentially increase the overall cost, as the value attributed to the barter transactions could be higher than the actual value of those transactions, increasing the VAT that the universities would have to pay.

6.4.4 YHMAN facilitates the mutual provision of service between universities, covered by a Memorandum of Understanding Our understanding is that there will be two strands to the overall arrangement. Firstly YHMAN/Data will buy (lease) physical space from a host university and let the space to a user university on a needs basis. If this is all YHMAN/Data will do then it should be possible to view the leases/lettings to/by YHMAN/Data as exempt from VAT unless a host university leasing to YHMAN/Data has opted to tax the building containing the data centre space. Secondly the member universities would provide and take other services

55

YHMAN Shared Virtual Data Centre

Business Case Final

(‘the mutual services’) from each other, with all transactions being covered by MoUs and with only balancing payments changing hands. The mutual services would cover things such as IT/IS staff support, including helpdesk and management duties

The mutual service mainly will work broadly on a time spent basis i.e. if university A uses university B’s staff for 20 hrs and university B uses university A’s staff for 30 hrs then university A will charge university B for 10 hrs staff time i.e. the net time. HMRC may well view this as a barter transaction and in pure VAT technical terms HMRC may hold that university B should raise a VAT invoice for 20 hrs and university B for 30 hrs, both of which would be liable to VAT. Whether it will be possible to persuade HMRC to take a net view i.e. charge VAT on only £10 in the above example, is debateable but it is worth opening a dialogue with HMRC on this point. One other possibility might be to use the VAT exemption provided by article 132(f) of the EU VAT Directive currently being looked at by HEFCE with HMRC in respect of the North Eastern Universities Purchasing Consortium to obtain VAT exemption for these mutual services. However discussions with HMRC are not particularly advanced and agreement on the conditions for exemption under article 132(f) could be some way off. In order to stand a chance of obtaining exemption for these mutual services we believe it would be necessary for any charging to be no more than a sharing of costs i.e. no mark up, and the member universities would operate on a consortium basis i.e. as an independent group of universities rather than through a separate company. Clearly this is an added complication and needs careful consideration given the likely delay in obtaining any agreement from HMRC to the article 132(f) exemption. However as the result would be to avoid creating a VAT cost on what will mainly be staff time this possibility may be worth considering if it is operationally feasible. We do not recommend this route for SVDC service provision, as the potential complexity of implementing the model, and the restriction on what services could be offered and how. In addition, from a corporate tax perspective HMRC are likely to impute a market value on the transactions under the transfer pricing rules. Establishing the MoUs and monitoring performance of service provision would be a relatively complex exercise. The issue of service performance also comes in here, with mutual provision of service offering the potential for uneven quality in service provision. This could quickly lose the trust of the Joint Venture universities in the service. This model could however be advantageous to groups of universities who do not already collaborate within a company structure, like the YHMAN Joint Venture universities do. Our recommended option from a VAT, wider tax and service provision perspective is 6.4.1.

56

YHMAN Shared Virtual Data Centre

Business Case Final

6.5

Facilities Management Arrangements The YHMAN personnel are envisaged as falling into two groups, the management (and Administration) staff, and the technical staff. The Management staff will be located in the existing YHMAN office facilities. The technical staff will be a distributed team employed directly by the institutions providing particular data centre space to YHMAN. This will allow YHMAN business to be conducted by technicians who are locally based and understand the physical infrastructure in which the equipment is hosted. If specialist input is required to the hardware itself, the institution to whom that hardware belongs would be expected to provide that input as they do currently. For all YHMAN work the technicians carry out, YHMAN will be invoiced by the providing institution, as currently happens with the YHMAN staff employed by Leeds university.

6.6

Agreement on Financing of Shared Services Vehicle There are different options open to the financing of the SVDC project and to a certain degree this depends on how deeply integrated a service the Joint Venture universities wish to provide. If they wish to pool their data centre requirements and provision within YHMAN and commission YHMAN to build and provide data centre space, then they will need to provide at least a proportion of the capital they would have spent on new facilities to YHMAN to allow it to pool this capital and take advantage of economies of scale. If they simply wish to maximise the use of existing data centre space, then direct capitalisation will not be necessary, as the revenue income will be sufficient for YHMAN to proceed. One option that is open to the YHMAN Board is to use their current reserves to finance the construction of a YHMAN data centre that can then realise revenue income to cover its costs.

57

YHMAN Shared Virtual Data Centre

Business Case Final

7

Conclusion This report provides a high level view of the potential for the SVDC to deliver benefits to the Joint Venture universities in terms of lower cost of procuring data centre services, and potentially better quality of those services. It was never intended to provide a detailed business case giving a fully considered view of the best way forward to deliver the maximum benefit. Rather it was intended to answer the question of whether the further investigation, at a much more detailed level, of the SVDC was worthwhile, or whether the benefits were too small to warrant that level of investigation. The document clearly shows that based on a standard set of assumptions there are potentially significant benefits to be achieved and that it is therefore worth moving to the next stage of the project. The next stage is to develop the Technical Proof of Concept currently underway to deliver some real technical insight into what some of the issues of the SVDC are going to be. In parallel to the TPoC we recommend agreeing with HEFCE the £150,000 funding necessary from them to combine with £65,000 of YHMAN money and facilitate the delivery of the Pathfinder and Joint Strategic Planning phases. These will develop the project to understand the cultural and process issues that could be encountered in a live service. It will also facilitate the Joint Venture universities working closely together to plan jointly how they are going to address the provision of commodity IT infrastructure requirements in the future. Out of this phase will come an agreed set of requirements for the data centre specification and management, a set of service levels that the SVDC will be expected to meet, and an understanding of how the institutions and YHMAN can work together effectively to deliver this project, along with a more detailed view of the costs and benefits of carrying out the project. We recommend progressing to this phase as soon as possible by agreeing the funding from HEFCE and YHMAN, and gaining agreement from Joint Venture universities on who the customer and provider universities will be for the Pathfinder, and what form the Joint Strategic Planning will take. This is a critical part of the project and we recommend planning this into diaries early so that work can get started on it, and a timetable of action with agreed milestones put in place.

58

YHMAN Shared Virtual Data Centre

Business Case Final

8

Appendix I – Costs of Data Centre provision

59

YHMAN Shared Virtual Data Centre

Business Case Final

60