The customer is always
WEBINAR: On-demand webcast
Next-Generation Applications Require the Power and Performance of Next-Generation Workstations REGISTER >
In an effort to capture a growing percentage of customers, Geiver and Erkonen decided in 2001 to use SAN technology to develop an analytics effort that would help analysts determine whether to offer credit to new customers.
After an ROI survey conducted by Hewlett-Packard Co. reported an estimated $3.5 million returnor about 592 percenton the deployment of a SAN, Premier Bankcard implemented a 30GB data warehouse using an HP ProLiant 3000 server that drew data stored on an HP StorageWorks EMA (Enterprise Modular Array) 12000 storage array. By using SAN technologies, Premier Bankcard could centralize its storage, allowing the company to speed access to information and improve the productivity of its data warehouse.
"We knew we were doing things right because we had to write a business case to prove we could get huge returns from a centralized storage strategy," Geiver said. "The potential ROI was just phenomenal and actually ended up exceeding our expectations."
As employees increasingly relied on the data warehouse to generate reports on customers, Erkonen and Geiver began a data warehouse expansion in 2002. Premier Bankcard customers regularly use their credit cards for such day-to-day purchases as groceries, which meant the amount of data generated and stored was growing exponentially.
At the same time, Premier Bankcard was expanding internally and adding more analysts. The organization began seeing file loads several hundred megabytes in size per day. To support this growth and level of activity, the company added a second HP StorageWorks EMA array in March 2003.
At the time, users pulled information from the data warehouse into Microsoft Corp.'s Excel and Access, as well as Business Objects S.A.'s Crystal Reports; users worked on the files using file shares on the ProLiant server.
Over time, users began to extract massive amounts of datasome users were pulling as many as 1 million rows of data off the server and into an Excel spreadsheet. A bottleneck developed, with some queries taking as long as 22 hours, Erkonen said.
In an attempt to resolve performance issues, Premier Bankcard deployed Microsoft's SQL Server 2000 database, but greater data growth followed. When the SQL Server database reindexed, for example, it would increase by as much as 30GB overnight. Finding their queries backlogged, Premier Bankcard analysts tried to work around the problem by loading queries at night and over the weekend, which had a negative impact on scheduled backup processes.
Next Page: The price of success.