Thom King, IST–DS
UC Berkeley comes a step closer this month to having a robust enterprise data warehouse to support decision making and related information needs across the campus. This month, Data Services will publish the Data Warehouse Architecture, the second major deliverable from the Data Warehouse Roadmap project, a joint project of IST and the campus Data Stewardship Council.
The Data Warehouse Architecture describes the implementation design of an enterprise data warehouse for the campus. The Architecture builds on the Data Warehouse Business Requirements Document, published several months ago as the first deliverable of this two-phase project.
A data warehouse is a computer-based set of facilities for delivering information to support understanding and decision making. The information comes mostly from operational computer systems, such as the course enrollment or financial systems. The data warehouse organizes the information so that it is accurate, but also easy to understand and consistent over time — so that it's easy, for example, to compare what happened yesterday or last month with what happened a year or several years ago. In addition to well-organized information, a data warehouse provides tools to help users find information and understand and evaluate it. Some of these tools are predefined reports that select relevant information and organize it for understanding. Others are query and analysis tools that help analysts discover patterns and draw conclusions from the information.
What makes an enterprise data warehouse? Its scope.
Building an enterprise data warehouse is a big job, and it doesn't happen all at once. Instead, they are usually built a step at a time over several years, with most-needed data collected first and other data added over time. It takes a good plan and lots of teamwork from experts from across the campus. Still, most of our peer universities have an enterprise data warehouse under development. Some, such as MIT, have mature data warehouses combining many data subject areas. Other universities are just getting started.
The Data Warehouse Roadmap project began in fall 2005 in the Data Stewardship Council (DSC), which is a cross-campus group of data proprietors and data custodians charged with helping the campus make the most of its information assets. Many members of the council — indeed, many leaders from across the campus — felt that the campus would benefit from migrating its several existing reporting systems to an enterprise data warehouse. This would build on success the campus experienced with systems such as the BAIRS reporting system, which uses data warehouse technology to deliver reports about finance and human resources. Knowing that building an enterprise data warehouse is a big effort that can be tackled many different ways, the Council wanted to have a development plan that was driven by what the campus needed most and most urgently. And so the roadmap project was born.
With business sponsorship from then VC–Budget & Finance William Webster and from Shel Waggener (then director of IST–CCS, now CIO), the DSC and IST put together a joint project to make a needs-driven plan for implementing the EDW. It was a two-stage project. In the first stage, a team of analysts led by Helen Norris and Dennis Hengstler, cochairs of the DSC, interviewed functional users from across the campus to understand both campus information needs and important success criteria for the warehouse. The team talked with researchers, course planners, administrators, Deans, and Vice Chancellors
The study identified a dozen significant ways an EDW would benefit the campus, and also found that broad campus participation and executive support would be needed to realize it. With these results, the sponsors are working with other members of the campus executive management to establish priorities and funding for implementation.
In the second phase of the project, leadership shifted to the IT members of the project, though analysts from the first phase continued to participate. Thom King, data architect, and the Data Services data warehouse team worked with a team of analysts and designers from across IST to design the implementation architecture for the warehouse. They designed the data structures, applications, tools, and security mechanisms needed to realize the business opportunities outlined in the requirements study. Having a new Data Services unit in IST definitely helped move the effort along.
Comprehensive data architecture is a key component of the new architecture. A grid structure that links data from across functional areas with a set of common context information will allow the campus to add data one increment at a time and still be able to combine that data for the kinds of integrated analysis that only an EDW can provide. In addition, the architecture provides some new security mechanisms and some new planning and analysis applications.
At a time when the campus is facing a broad set of changes and challenges, having a robust enterprise data warehouse to help understand and make choices should pay off now and for years to come.
[ iNews | Search | IST | UC Berkeley Computing | UC Berkeley ]
iNews: UC Berkeley information technology news channels
Copyright 2006, The Regents of the University of California