Get the Most out of Data on the HP e3000
When HP's Commercial Systems Division (CSY) dropped an "e" in front of the HP 3000, they told the world that the platform is a player in e-commerce and e-services, as well as data warehousing and business intelligence. Find out if they really meant it.
When HP’s Commercial Systems Division (CSY) dropped an "e" in the HP 3000, rebranding it the HP e3000, CSY was telling the world that the platform is a player in e-commerce and e-services. That includes data warehousing and business intelligence.
"Many people don’t know that," says Alvina Nishimoto, CSY’s R&D Program Manager. "We hear people say, ‘Oh, it’s so hard to get the data I need out of my IMAGE database.’ But it has nothing to do with the IMAGE database. It has to do with the fact that our applications have been very highly tuned for online transaction processing, and the data is organized as such. It’s not well organized for reporting and the trending type of information people want from a business intelligence database."
In other words, production systems are organized to optimize online transaction processing, but business intelligence requires different types of organization (see Table 1). This is where tools come in. They’re used to optimize data for reporting purposes.
Data processing, support of business operations
Process oriented, current values, highly detailed
Subject oriented, current and historical values, summarized and sometimes detailed
Ad hoc, some repetitive reports and structured applications
Data entry, batch, OLTP
End user initiated queries
"A lot of what customers want to do with data is perform queries that cannot be predicted ahead of time," Nishimoto says. "Take, for example, buying trends. Businesses want to be able to look at one thing, think that’s interesting, and try a different query.
Well, the majority of databases have to predefine these queries in order to get good performance out of their e-intelligence solution, so that kind of negates the benefit. That’s why we emphasize tools – because customers are going to do ad hoc queries and do them on a large amount of data. And tools – not hardware – provide for this."
"This has nothing to do with IMAGE, specifically," Nishimoto explains. "Relational databases have similar issues with business intelligence and data warehouses, and so they require similar sorts of tools. Our tool vendor partners, like Taurus and DISC, sell their products for other kinds of databases as well."
CSY has had tool vendors as partners for some time because HP e3000 users have demanded business intelligence capabilities. Take, for example, users of Smith-Gardner’s mail order and cataloging system. "These users have been running WebOrder [Smith-Gardner’s Web order fulfillment system] for quite a while," Nishimoto says. "They look for trends so they can get ideas, special promotions and so on that can increase sales."
DECISION VAULT: BI ON THE HP e3000
CSY offers a business intelligence solution for the HP e3000 that allows users to remain with the reliable and familiar IMAGE database and still obtain the reporting and trending capabilities they need. The bundle, called Decision Vault, involves four HP partners: DISC, Taurus Software, Quest Software and Brio Technology.
For the solution, Taurus and Quest jointly developed BridgeWare. BridgeWare performs an initial load of the data warehouse and carries out the essential data transformation, data scrubbing, data denormalization, and loading of data. BridgeWare updates through Quest’s Netbase (SharePlex/iX), which detects changes automatically as they occur.
After the data has been loaded into the IMAGE warehouse, it is queried, analyzed, and retrieved by end users through DISC’s OMNIDEX. OMNIDEX is a database search engine that supplements the existing database, supplanting the native database indexes with one specifically designed for high-performance, multidimensional access and dynamic data aggregations. It provides additional access not possible with the native IMAGE database alone. DISC’s indexing capability boasts impressive performance, as demonstrated by a test of an operational healthcare IMAGE database containing 9 million records and running on an HP 3000/967 in a batch-reporting scenario. In the test, complex queries were executed in 21 to 50 seconds.
OMNIDEX integrates transparently with the ODBC analysis/reporting tool BrioQuery. BrioQuery, designed for end users, provides an intuitive query, analysis and reporting interface to the data warehouse.
Decision Vault provides a native business intelligence solution for the HP e3000. It is used especially by businesses in CSY’s targeted vertical markets, including healthcare (Amisys), mail catalog (Smith-Gardner) and manufacturing (MANMAN).
– Jean Nattkemper is the Editor at Large for HP Professional. She can be reached at firstname.lastname@example.org.