Tuning up the warehouse

As government agencies cram their data warehouses with more information, federal information technology managers need to ensure those important decisionsupport systems don't get bogged down.

As government agencies cram their data warehouses with more information,

federal information technology managers need to ensure those important decision-support

systems don't get bogged down.

A new product from SAS Institute Inc. will help IT managers better understand

what users want from agency data warehouses and how efficient the warehouses

are at providing the correct information. That information can help managers

fine-tune the data warehouse and decide when system upgrades are necessary.

A data warehouse is a repository of information from multiple sources

and is structured for querying and reporting. SAS' Intelligent Data Warehouse

Manager, introduced last month, provides the warehouse administrator with

in-formation on what database tables users are looking for, what information

was retrieved and how long it took to retrieve it, said Frank Lieble, program

manager for IT performance solutions at SAS Public Sector.

"It addresses the performance of the infrastructure itself," Lieble

said. "Data warehouses are growing so fast that the tools needed to manage

them are in great need."

Because of the high cost of managing data warehouses, a tool like the

Intelligent Data Warehouse Manager can help maximize the current investment

and reduce the cost of managing the systems, Lieble said. The average cost

of a data warehouse is $2.5 million, he said.

The tool will help the IT manager decide whether it is necessary to

buy new hardware or balance the data warehouse's request load more evenly,

he said. The software solution supports three Unix platforms: HP UX, IBM

RS/6000 and Sun Solaris. It can be used with data warehouses from Oracle

Corp., Informix Software Inc., Sybase Inc., NCR Corp. and IBM Corp., Lieble

said. About 75 percent of government data stores are Oracle, he added.

Oracle offers its own tools to monitor the performance of its Oracle

Warehouse Builder, said Steve Jones, business intelligence and data warehousing

sales consulting manager for Oracle Service Industries, the firm's government

group.

Oracle Enterprise Manager allows the administrator to schedule an automated

refresh of the data in the Warehouse Builder. Oracle Workflow has more sophisticated

options for maintenance. And Oracle Warehouse Builder's run-time sharing

captures information about the query process to show progress of any activity

within the warehouse builder.

It is important to monitor and audit a data warehouse, but the challenge

is taking the information and using it to fix the problems, Jones said.

"There's a lot of maintenance that goes into warehousing," Jones said.

"A warehouse is never complete. It's nice to have a tool because doing it

by hand can get out of hand."

Government agencies probably won't rush out and buy performance management

tools until their data warehouses have grown beyond a manageable size, said

Ray Ransom, computer specialist at the Centers for Disease Control and Prevention.

CDC is in the early stages of creating a Sexually Transmitted Disease

data warehouse derived from weekly reports from states, aggregate disease

reports, special reports produced at CDC, Census data for calculating rates

of disease and other sources.

By manually maintaining a data warehouse, the manager is aware of the

structure and can reload and repopulate data files as they are updated,

Ransom said.

When a data warehouse becomes too big to do that, a tool such as SAS'

would come in handy, Ransom said. In any data warehouse project, the issue

is whether to buy commercial off-the-shelf tools or develop them internally,

he said.

"The real power of these tools comes in when the load process is cycling

into the next business day. As [the data warehouse] grows, that administration

has to be maintained electronically," he said.

Many agencies are still in the early phases of data warehouse implementation,

so tools like SAS' are not yet a priority. "To me, it's the icing on the

cake," Ransom said. "It's a tool we'll get when we're up and running full-tilt."