Borrador V 0.9
1
Data Warehouse Engineering Process (DWEP) with U.M.L. 2.1.1. Edwar Javier Herrera Osorio,
[email protected] Universidad Nacional de Colombia
Abstract— This paper presents an update DWEP to version 2.0. DWEP in the use of use case diagrams, class diagrams, package diagrams, deployment diagrams. Is the use of the same with their updates, it also proposes the use of state diagrams, activity diagrams, composite diagrams, structure diagrams, interaction diagrams and overview diagrams Index Terms— Data warehouse, UML, Unified process, data models
I.
T
II.
RELATED WORK
Recent years have developed several methodologies for the development of data warehouses which defines the following levels of abstraction [7]: Conceptual, logical and physical. Conceptual Data Model: Represents the interactions between the entities and relationships. This model is closer to real world problems to solve. Highlights the following patterns in the data warehouse: Model Multidimensional / ER (Sapia) [8], model Star / ER (Tryfona) [9], GOLD model (Trujillo) [5, 10], model Husemann [11], YAM2 model [12].
INTRODUCTION
he data warehouse (DW) is one of the components of the intelligence business, Bill Inmon defines it: “... A data warehouse is a subject-oriented, integrated, timevariant, nonvolatile collection of data in support of management’s decisions...” [1], and Ralph Kimball: “… the Data Warehouse is a collection of data in the form of a database that stores and organizes information that is extracted directly from operational systems (sales, production, finance, marketing, etc..) and external data…”[2]. Building a DW is a challenging and complex task because a DW concerns many organizational units and can often involve many people. Lujan poses at the 2004 [3,4] Data Warehouse Engineering Process (DWEP), a methodology for building the data warehouse based on the Unified Modeling [5] and the Unified Process (UP) [6], which allows the user to tackle DW all design stages, from the operational data sources to the final implementation and including the definition of the ETL (Extraction, Transformation, and Loading) processes and the end users' requirements. The rest of the paper is structured as follows. In Section 2, we briefly present some of the most important related work and point out the main shortcomings. In Section 3, we summarize DWEP: first is presented phases, then workflows and your use diagrams based in UML version 2.1.1 (the results achieved so far) and shows the use of these devices in the workflows that make up our process. Finally, we present the main contributions and the future work in Section 4.
Logical data model: The objective of the logical data model is to describe in as much detail as possible, without considering how they will be physically in the database. Is this model includes entities, relationships and their interaction, the data types of all attributes of each entity, the definition of primary and foreign keys, definition of the extraction, transformation and loading (ETL), among other activities. Physical Data Model: The physical data model includes all the specification of all tables and columns, following the business rules to determine the design of the data warehouse. In this model, you write the code to create tables, views, integrity rules, multidimensionality consultations. On the other hand are the different methodologies for the development of data warehouses [3, 5, 13, 14, 15, and 16], most shortcomings: do not include a visual modeling language, not to propose a series of steps or phases, or based on an application (for example, the star diagram of relational databases). In 2005, Lujan proposed a methodology based on the Unified Process (Data Warehouse Engineering Process DWEP), which is based on UML version 1.4. The DWEP propose a collection of artifacts for standardization. In conclusion DWEP claim upgrade to version 2.1.1. of UML which gives us more devices to implement the data warehouse.
Borrador V 0.9
2
mitigating the risk of technological exploration of the programming language in terms of user interface is concerned. For this first iteration was completed with a functional prototype for testing software and the definition of the model for implementing the user interface. Construction Phase: The construction phase starts from the baseline architecture that is specified in the design phase, and its purpose is to develop a product ready for initial operation at the end-user environment. Transition phase: Once the project enters the transition phase, the system has reached initial operating capability. This phase seeks to introduce the product in its operating environment. Workflows DWEP
Figure 1 DWEP [5]
III.
DATA WAREHOUSE ENGINEERING PROCESS
Lujan in his doctoral thesis [5] presents a Data Warehouse Engineering Process (DWEP) based on the unified process. The UP is a methodology for software development proposed by OMG [17], its main features are: it is iterative, is addressed by the use cases is based on stages of development, using UML as a graphical language models [18 and 19]. The UP and DWEP is composed of four phases [5 and 20]: inception, design, construction and transition (view Fig. 1). Phases UP and DWEP Inception Phase: This phase is to develop the project analysis to justify its implementation. To achieve this there is a general description of the project, a planning based on interactions of the phases, there are critical risks and establishes the basic functionality of the software architecture description of a candidate. Development phase: Once the initial phase is to build a robust architecture for building software. This phase seeks to establish the rationale for implementing the use cases and artifacts of the final system component, in addition to
In general terms the UP, workflow is a set of activities in a given area resulting in the construction of artifacts (a text, a diagram, a web page, code in programming language, etc.). DWEP in version 2.1.1 present 20 diagrams (5 process and 3 levels), view table 1.This diagrams is use in the different workflows. Requirement: During this workflow, end users specify the measures and add more interesting, dimensional analysis, queries used to generate periodic reports and frequency of updating the data. For the development this stages the UP use of use cases. View Figure 2. This helps to understand the system and the requirements and functions for the solution. Furthermore, it must be like the interactions of the system. Analysis: The purpose of this workflow is to improve the structure and requirements from the requirements stage. This step documents the incumbent systems that feed the data warehouse. The unified process diagram of the proposed use of class diagrams, objects, communications, and deployment. DWEP proposed use the Source Conceptual Schema (SCS, View 3), Source Conceptual Object Schema (SCOS, View 4), Source Logical Schema (SLS, View 5), Source Logical Comunications Schema (SLCS, View 6) y Source Physical Schema (SPS, View 7).
Borrador V 0.9
3 Source (S)
Integration
SCS (Class)
DM (Class)
SCOS (Object)
DWSS (Sequence)
Conceptual
DWSMS (State Machine)
Data Warehouse (DW)
Customization
Client (c)
DW CS (Class)
DM (Class)
CCS (Class)
DW LS (Class)
Exporting Process (Class) CLS (Class)
DW PS (Comp & Deployment)
Transportation (Deployment)
DWAS (Activity) SLS (Class) Logical
Physical
ETL (Class)
SLCS (Communication)
SPS (Comp & Deployment)
Transportation (Deployment)
Diagram
Diagram
CPS (Comp & Deployment)
Table 1 DWEP 2.1.1 Diagrams
Figure 2 Use Case diagrams [5]
Figure 3 Source Conceptual Schema [5] TV:Products
Miami:Cities
Sony:Customer
001:Orders
Radio:Products
Figure 5 Source Logical Schema :Cities Play Statio
1: Read_table
TV2:Products
2: Read_table
:Customer
002:Orders
Job System 3: Read_table Radio2:Products
4: Read T abl e
:Orders
Figure 4 Source Conceptual Objects Schema :Products
Figure 6 Source Logical Communications Schema
Borrador V 0.9
4
Open Source
DWSD Customer
Read and extract data to relati onal data base
Transform and l oad i n temporal Space in DW
Figure 7 Source Physical Schema
Load to temporal Space DW to DW
Figure 10 Data Warehouse State Machine Schema
Design: At the end of this workflow, the structure is defined in the data warehouse. The main result of this workflow is the conceptual model of the data warehouse. The UP proposes the use classes structured into packages, design of subsystems defined interfaces (components) and the form of collaboration between the classes. The DWEP proposes the use Data Warehouse Conceptual Schema (DWCS, View Figure 8), Client Conceptual Schema (CCS),el Data Mapping (DM, View Figure 9.), Data Warehouse State Machine Schema (DWMSS, View Figure 10.) y el Data Warehouse Activity Schema (DWAS, View Figure 11.).
Figure 11 Data Warehouse activity schema[21]
Implementation: During this workflow, the data warehouse is built: The physical structure of the data warehouse is built, start to receive data in computer systems operations, is tuned for optimized performance, among other tasks. The process proposed as unified engine components diagram. View figure 7. The DWEP propose use: Data Warehouse Physical Schema (DWPS, View Figure 12), Data Warehouse Logical Schema (DWLS, View Figure 13), , Client Logical Schema (CLS), Client Physical Schema (CPS), Data Warehouse Secuence Schema (DWSS, View Figure 14), ETL Process (View Figure 15).
Figure 8 Data Warehouse Conceptual Schema [5]
Figure 9 Data mapping [5]
Figure 12 Physical diagram of the data warehouse [5]
Borrador V 0.9
5 Workflows for maintenance and development post are not in the unified process and only part of the engineering process of the data warehouse. Maintenance: Unlike most systems, the data warehouse is a process that feeds constantly. The purpose of this workflow is to define the loading and updating processes necessary to maintain the data warehouse. This workflow starts when building the data warehouse and is delivered to end users, but does not have an end date. During this study, end users may have new needs, such as new downloads, which triggers the beginning of a new iteration with the requirements of workflow.
Figure 13 Logical diagram of the data warehouse [5]
Relacional DB:Customer
DWTem poralSpace:Customer
DW:Customer
Revisions post development: This is not a workflow of development activities, but a review process to improve future projects. If we keep track of time and effort invested in each stage is useful in estimating time and needs to generate the requirements for future developments.
Sales manager
IV.
extract(Parameter)
CONCLUSION
Transform(Parameter)
Load(Parameter)
Figure 14 Data Warehouse Secuence Schema
Developing a DW is a complex, expensive, time consuming, and prone to fail task. Different DW models and methods have been presented during the last few years. However, none of them addresses the whole development process in an integrated manner. In this documents, we have presented our DWEP based on the UML and the UP, which addresses the analysis and design of both the DW back-stage and front-end. For this task, we have extended the UML in order to accurately represent the different parts and properties of a DW. Following our approach, we design a DW as follows: We use UML to model the data sources of the DW at the conceptual level. The use of the same notation (UML) for designing the different DW models and the corresponding transformations in an integrated manner.
Figure 15 ETL Process [5]
Tests: The aim of this work is to verify the application to work correctly. More specifically, the effects of the tests are: Planning the evidence needed to design and implement the tests by creating test cases and perform tests and analyze results of each test.
The use of the UML importing mechanism, which guarantees the designer that each element is defined once, because the same element can be used in different models. REFERENCES [1] W. Inmon, Building the data warehouse. Wiley, 2002. [2] R. Kimball and M. Ross, The Data Warehouse Toolkit: The Complete Guide to Dimensional Modeling. Wiley, 2002.
Borrador V 0.9 [3] S Lujan and J Trujillo. A Data Warehouse Engineering Process. Advances in Information Systems, Springer Berlin / Heidelberg, Volume 3261/2005 , pp. 14–23. [4] S Lujan , Data WareHouse Desig with UML, PHD. Thesis, Universidad de Alicante, 2005 [5] Object Management Group (OMG): Unified Modeling Language Specification 2.0, Internet: http://www.omg.org/technology/ documents/ modeling_spec_catalog.htm#UML. 2009 . [6] Jacobson, I., Booch, G., Rumbaugh, J.: The Unified Software Development Process. Object Technology Series. Addison-Wesley . 1999. [7] Steel,T.B.,Jr. (Chairman): ANSI/X3/SPARC Study Group on Data Base Management Systems Interim Report; ACM SIGMOD FDT, Vol. 7, No. 2, 1975. [8] C. Sapia, M. Blaschka, G. Hofling, and B. Dinter. Extending the E/R Model for the Multidimensional Paradigm. In Proceeding of the 1ST International Workshop on Data Warehouse and Data Mining (DWDM’98), volumen 1552 of Lecture Notes in computer Science, pages 105-116, Singapore, November 19- 20 199. Springer- Velang. [9] N. Tryfona. F. Busborg, and J.G. Christiansen. starER: A Conceptual Model for Data Warehouse Desing. In proceedings of the ACM 2nd international Workshop on Data Warehousing and OLAP (DOLAP`99), pages 3-8, Kansas City, USA, November 6 1999. ACM. [10] J. Trujillo. The GOLD model: An Object Oriented multidimensional data model for multidimensional database, Symposium on Applied Computing Proceedings of the 2000 ACM, symposium on Applied computing- Volume 1, Italy, pages 346-350, 2000. ACM. [11] B. Husemann, J. Lechtenborger, G. Vossen, Conceptual Data Warehouse Desing, Proceeding of the International Workshop on Design and Management of Data Warehouses (DMDW’2000), StockHolm, Sweden. [12] A. Abello, J. Samos, and F. Saltor. YAM2 (Yet Another Multidimensionañ Model): An extension of UML. In International database Engineering applications Symposium (IDEAS’02), pages 172-181, Edmoton Canada, July 17-19 2002. IEEE Computer Society. [13] Kimball, R.: The Data Warehouse Toolkit. John Wiley & Sons (1996) (Last edition: 2nd edition, John Wiley & Sons, 2002). [14] Giovinazzo, W.: Object-Oriented Data Warehouse Design. Building a star schema. Prentice-Hall, New Jersey, USA (2000) [15] Cavero, J., Piattini, M., Marcos, E.: MIDEA: A Multidimensional DataWarehouse Methodology. In: Proc. of the 3rd Intl. Conf. on Enterprise Information Systems (ICEIS’01), Setubal, Portugal (2001) 138–144 [16] Moody, D., Kortink, M.: From Enterprise Models to Dimensional Models: A Methodology for Data Warehouse and Data Mart Design. In: Proc. of the 3rd Intl. Workshop on Design and Management of Data Warehouses (DMDW’01), Interlaken, Switzerland (2001) 1–10
6 [17] [21] Object Management Group (OMG). Unifie Modeling Language (UML), version 2.0, consultado marzo de 2008 Internet: http://www.uml.org/ [18] Booch Grady, Rumbaugh Jim, Jacobson Ivar, “UML, El lenguaje unificado de modelado”, consultado en internet http://www.itescam.edu.mx/principal/sylabus/ fpdb/recursos/r25380.PDF [19] Fuentes Lidia, Vallecillo Antonio. “Una Introducción a los Perfiles UML, Consultado en Internet” http://www.lcc.uma.es/~av/Publicaciones/04/ UMLProfiles-Novatica04.pdf. [20] Jacobson, Ivar; Booch, Grady; Rumbaugh, James. “El proceso unificado de desarrollo de software.”, Addison Wesley. Madrid, ES. 2000. 438 p [21] Veronika Stefanov, Beate List, Birgit Korherr. “Extending UML 2 Activity Diagrams withc Business Intelligence Objects” First A. Author Edwar Javier Herrera Osorio (05/10/1977), systems engineer, Universidad Distrital 2004, specialist in database development, foundation university of Bogotá Jorge Tadeo Lozano, 2007, Master candidate in systems engineering and computer 2008, Universidad de Colombia.