PHH Lets Clients "Take the Wheel" on Their Data

Software Magazine, August, 2001 by Vance McCarthy

PHH ARVAL, A LEADING PROVIDER and manager of the "company car" to thousands of North American employees, is making good on an ambitious vision to marry data integration and the Internet. Through its PHH InterActive B2B Web site, the fleet management solutions company is providing execs at Fortune 500 firms (and thousands of other businesses) unprecedented access to its billing and activities records. In the process, the PHH InterActive system is helping executives at client firms keep alive that best-known (and most used) company perk--the company car.

The PHH Interactive business intelligence system provides PHH's partners and customers a custom GUI to a multiterabyte database. The central database is in fact a massive data warehouse (940 GB and growing at about 18 GB per month) along with a data mart (200 GB and growing at about 1.5 GB per month). The entire data warehouse comprises multiple data sources that track over half a million leased vehicles through their lease, operations, and repair, says Jeff Wolf, PHH InterActive's project manager.

"Today we've basically got a business intelligence system that provides our staff, clients, and supplier partners with Web-based access for queries or analysis on their [leased vehicles] accounts," Wolf says. The system allows clients to check a variety of data points on their fleet's history (including, vehicle orders, fuel purchases, accident and repair information, operational trends, and more). Supplier partners such as automotive service and repair shops can get updates on pending actions (including finalizing and approving repair totals, submitting invoices, etc.). Over 4,000 client users utilize PHH InterActive regularly to help them effectively manage their fleets. The company's newest application, PHH InterActive for Suppliers, has some 115 suppliers signed up to use it right out of the gate.

Importance of Data Warehouse

The data warehouse is key to PHH InterActive, Wolf says. The data warehouse is made up of a variety of OS/390 files (flat files, datacom files) and Sybase source files, which have been targeted to a Sybase relational database running Sun Solaris.

To build the integrated data warehouse, PHH worked with integration tools company Evolutionary Technologies International (ETI) in Austin, Texas.

The biggest challenge to developing such a flexible data warehouse was developing both the initial "load" program as well as the update programs for daily and monthly data refreshment, says Wolf. PHH Arval used ETT*Extract tools to collect, process, and move the mainframe data to the Sybase target. As part of the solution, ETI's tools also generated close to 500,000 lines of necessary code (in both Cobol and C) in six to eight weeks.

Tools Plus Team

In addition to choosing the right tools to build the data warehouse, PHH put together a project team that included business analysts as well as technical people, in order to better capture the requirements of its customers. "It was very important to make sure that the data warehouse we ended up with would match the kinds of queries and requests that most customers would want to access, and so business analysts would help keep us on track," says Wolf.

Another important aspect was the concept of "timeboxing," Wolf added. Timeboxing is a technique developed over the last several years by ERP consultants to keep potentially massive projects from becoming resource-draining efforts that simply get out of hand, and never seem to end. Under disciplined approaches for the "timeboxing" method, a large project needs to be segmented into manageable modules; maximum productivity is derived when the (sub)project length is set at six months and team size is at a maximum of eight to 10 people.

The PHH business analysts and IT staff collaborated to develop their own version of the "timebox" for their projects.

To make sure the target platform would be a multifunctional data warehouse (capable of accommodating a variety of queries from various partners and/or clients), PHH Arval applied business rules to transform original mainframe data fields, which contained important variables such as dates, coded information, and financial data. Moving the data from the multiple mainframe sources to the Sybase target required nearly one million lines of code and hundreds of batch jobs, which loaded Sybase with almost 100 database tables, relying on core data and metadata entries.

This attention to detail at the outset means that as the data warehouse needs updating or scaling to include new data, such revisions can be made quickly, Wolf says. Depending on the complexity of the change, the goal was to implement revisions in one to five days. And that's important to PHH InterActive's future. "What's next for us is to look at how we include materials from some of our partners in our data warehouse," he says. "We're just in the beginning phase of scoping that out," Wolf says, "but if we didn't have a lot of technology that would scale and that we could reuse, I don't think we'd be able to even consider it."