Get the Complete Picture

The data architect plays one of the most critical -- yet misunderstood -- roles on the e-business development team

By Rajan Chandras

Continued from Page 1

Multifarious Duties

Business insight. The data architect must work closely with the client, external agents (such as external data providers), and the business team to fully understand and represent the business model. "Data discovery'" is particularly important: that is, discovering the various sources of data that will come into the application architecture, and then determining the "what, how, when, where" of this data. Early in the project, it may be impossible to determine exactly what external data will be needed and how it will be acquired; the client or business may not have this knowledge up front. In the classic venture capital-financed project model, the clients may have the money to fund the software development even before they have determined the details of their business model. Thus, the data architect will have to work very closely with the other members of the architecture team and the business team to figure out the detailed business model and translate it into a feasible and acceptable technical solution.

TO PROTECT AND TO SERVE

Data architects are valuable assets at every project stage Initiation and Planning Data architecture scoping Data architecture planning Recruiting Data Discovery Design Logical data modeling Physical database design Database integration Object/relational modeling Technical specifications Support and validation External data interfaces Architecture and design ETL process architecture and design Development and Test ETL process development Data model maintenance Configuration management Development team support Test team support Deployment Deployment planning Deployment execution Deployment hand-over All Phases Project planning Team lead Interaction with client/business Interaction with external data providers

Technical expertise. Much of the data architect's role is technical in nature. Although data architects are not required to have the numerous and complex hands-on skills of a DBA, they must have a firm grounding in database and SQL concepts and have the ability to apply them to determine the optimal technical solution to a business requirement. It also helps for the data architects to know Java. They must have strong data modeling skills and previous experience with data warehousing and data ETL processes. The data architect should also be familiar with object-oriented techniques and object modeling. This experience will help the data architect work on creating the object/relational interface required by Java components when accessing relational data.

An understanding of XML will prove valuable to the data architect in enterprise application integration (EAI) and data integration across disparate systems and data sources. Specialized requirements of Web sites may also need the data architect to have (or gain) expertise on technologies for wireless, streaming data, and so forth. A key part of the data architecture is to design and plan for data distribution and replication, database performance, data recoverability, and database server fault tolerance. The data architect, along with the technical architect, must also work with vendors to evaluate and acquire database hardware and software.

Project planning. The data architect will be the de facto owner of the data architecture section of the project plan and must be familiar with project estimation, resourcing, planning, and scheduling.

You need a data architect on your team right at the beginning of the project. This is a critical point: Without a data architect involved in the project planning, there will be no ownership of any data- related activities. As a result, future data architecture activities may not be included in the project scope; instead, they will probably not even be identified, and resourcing for data architecture activities (data movement process development and testing, for example) will not be specified. You may be completely missing a significant piece of the project plan, and you know what happens in that case.

Team leadership. The data architect is the team leader for the DBAs and the ETL process developers and testers. Thus the data architect must have team leading experience, people management skills, and be a competent representative of the team to project management and other teams and team members. The data architect is also responsible for recruiting team members, so recruiting and interviewing skills are also required.

Resourcing generally begins with the database administrator. For any nontrivial dot-com project, being a DBA is a full-time job and requires an experienced professional under the direction of the data architect. The primary task of the DBA is to administer the databases. And this task is not trivial.

The typical dot-com project involves multiple databases: the development and test databases in the initial phases, and staging and production databases in the later stages. These databases reside on various computers ("boxes") and various operating systems. Managing all these database instances, supporting all the database users (developers, testers, and so on), data interfaces, and data movements will require all of the DBA's available skills and time. As an aside, of all the positions listed by recruiters, the "database administrator" must surely be one of the most misrepresented. I've seen recruitment advertisements that imply that any activity involving data must require a database administrator.

Rate This Article Rating: 1 (best) 2 3 4 5 (worst) Comments: Optional e-mail address:

Technical support. As the dot-com project spirals furiously toward the go-live date, the data architect has an essential responsibility of providing technical support on data and database issues to the development and test team. During the design phase, the technical or programming specifications will begin to roll out, and the data architect (in the avatar of the data modeler) will be involved in helping the developers understand the data model, flows and interfaces, as well as validating the technical specifications prior to their release. In the development phase, the data architect will be involved in resolving any database-related issues and data model queries. Simultaneously, the test team may have begun preparations for testing, and this again requires technical data-related support from the data architect and the database administrator.

Take Care of Your Data

• Bring the data architect on board as early as possible -- during the project initiation and planning phase.
• Ensure that data architecture tasks are an integral part of the project plan.
• Identify and assess external data sources as well as internal and external data interfaces early in the project.
• Allocate the resources (for example, time, people, hardware, and software) to build the data interfaces and data movement components.