San Diego Gas & Electric (SDG&E) is a major energy supplier to San Diego and southern Orange counties. SDG&E prides itself in offering innovative service to customers--features such as the company's seven-day, 24- hour toll-free hotline have earned SDG&E 95 percent customer satisfaction rankings for the past five years.

Implementation Summary

The specter of deregulation, due to go into effect in 1998, is motivating the entire utilities industry to find innovative ways of maintaining and expanding market share. For SDG&E, this meant investigating, prototyping and deploying decision support data marts to serve the company's various divisions. The first, a proof-of-concept data mart for the marketing division, was completed in October 1996.

SDG&E's overall goal was to create a number of data marts over time that can help the organization thrive in the face of new competitive dynamics. SDG&E selected Informatica's PowerMart's suite in February 1996, along with several other data manipulation products, to begin the proof-of-concept data mart.

"We wanted to get it all set up at the same time," says Wiltshire, "so we brought in all the new software in conjunction with our marketing data mart project." The marketing data mart's initial objective was to integrate data from three independent sources-- commercial customer consumption data, marketing program data and customer contact data. Among other things, this would help evaluate the effectiveness of SDG&E's demand-side management, or energy efficiency programs, to commercial customers.

In recent years, the Public Utilities Commission has issued rewards to utilities and their customers for reducing energy consumption. As a result, marketing efforts--to sell new, high-efficiency products, for instance--have grown. Now, with deregulation looming, the ability to identify and market to new users has become even more critical.

"This data mart gives our marketers the ability to drill down to a fine level of detail very quickly," says Wiltshire. "Before the mart, their queries were limited in scope: they could only look into a database of customers already participating in the programs. And they were hampered by time. It would take several weeks of sorts and merges to get the same level of response they can now get with a few queries of the data mart. The scope is improved, too. They can go out and ask the data mart, 'What commercial customers who used over 500 kilowatts per month over the past two years are not participating in a program?'"

SDG&E is now planning to broaden its use of data mart technology. Wiltshire and his management team are looking at ways to broaden the scope of the marketing mart-- taking it to residential customers next. Also, they are beginning another data mart in the fuel and powers division.

Just getting underway, too, is a program for making data mart and other information available via the utility's intranet, according to Wiltshire. "We're finding many classes of users here, and they all need different views of the data," he says. "We've got users who want canned reports and we've got business department analysts who want to get in and really slice and dice the information."

"That's what our data mart strategy is all about: to make it easy for all types of users to get at the data in whatever form they want. This will help extend the mart's reach to new users, and it will attract new, yet unthought-of queries--another benefit of the data mart concept," Wiltshire says.

In SDG&E's current configuration, PowerMart extracts source information from several databases. Two Microsoft SQL Server databases hold customer contact and marketing-program information, while IDMS and DB2 databases deliver commercial-customer energy utilization and other information. The IDMS and DB2 databases feed into a mainframe-based relational data mapping product, which also allows access to SDG&E's VSAM files. From there they were initially configured to go into third-party middleware-connectivity software, running on a dual-Pentium gateway server, where they would be joined by the SQL Server data.

The middleware product was supposed to join tables from these disparate systems in preparation for feeding the data mart. But in operation, the middleware was painfully slow, according to Wiltshire. "It took about eight hours to join two tables, one a mainframe table of about 86,000 rows, and the other a server table of 27k rows," he says. "The vendor came in, but was unable to tune the system to do any better. So we gave the job directly to PowerMart. We staged the 86k table in the data mart, and then went over, got the SQL Server table, joined it to the staged information --did the complete load--all in 24 minutes."


By employing PowerMart, SDG&E was able to:

  • Save development time and programming resources--PowerMart facilitates the process of warehouse design, and automates data mapping so users don't have to write COBOL extract programs or SQL query statements. This helped cut months off development time, and it eliminated the need to hire programmers.
  • Reduce system complexity--Adapting to change is an essential quality of a data mart, particularly in a proof-of-concept application. PowerMart made it easy to modify and recreate data mappings without having to recompile programs. Also, PowerMart comes with a repository, and populates it automatically with system meta data; with COBOL generators, writing and updating documentation would be a tedious process.
  • Increase system management efficiency--PowerMart employs comprehensive load-features that make it easy to set up, schedule and monitor data loads. While SDG&E is anticipating increased scheduler flexibility in the next release, the PowerMart system management features were a selling point because they free up data mart administrators from routine chores.
  • Streamline source extracts--PowerMart's wide source compatibility and powerful engine-based technology is helping SDG&E speed data extracts from Microsoft SQL Server and mainframe database sources.

Practical Advice

Establishing early agreement on business goals and committing the appropriate organizational resources can have as much of an impact on the success of a major data warehousing project as the technology involved. Given the importance placed on the bottom line, achieving buy-in from users and upper management on such resource-intensive projects is crucial. The most effective road to success: deploying strategic, distributed data marts building towards an enterprise data warehouse "from the bottom-up." Data marts provide a means to start small with data warehousing, systematically targeting important sectors of corporate data to increase revenue and maximize investment. Once in place, the stand-alone data marts can be synched up through integrated meta data for a full-scale, enterprise-wide data warehouse.

Register or login for access to this item and much more

All Information Management content is archived after seven days.

Community members receive:
  • All recent and archived articles
  • Conference offers and updates
  • A full menu of enewsletter options
  • Web seminars, white papers, ebooks

Don't have an account? Register for Free Unlimited Access