IBM Posts Data Analysis Record at Teraplex Integration Center
ACTION ITEMS

In a demonstration that proved the flexibility and scalability of the server for data warehousing, IBM analyzed what it claimed was the largest file ever created on an AS/400e, at its Teraplex Integration Center in Rochester, Minn. All 10 billion rows of data in the 2.5 terabyte file were examined in just 77 minutes, the equivalent of reading all the information in a stack of paper 75 miles high in under an hour and a half.

Coupled with IBM 1997 sales figures that show more than 3,000 AS/400s were implemented specifically for business intelligence purposes, this record reinforces the platform's ability to scale to analyze larger quantities of data and look at more variables than ever before, making it possible for companies to solve a wider variety of business problems. For example, a retailer that previously could analyze revenue by store or by product grouping only, would now be able to achieve greater precision by examining revenue by stock keeping unit (SKU).

Created using DB2/400 in a massively parallel processing (MPP) environment with symmetric multiprocessing (SMP) enabled nodes, the 2.5 terabyte file is the largest ever built on an AS/400. It was distributed across five AS/400e 12-way servers connected with OptiConnect fiber-optic lines. DB2/400 manages the 10 billion record file as a single database across all five servers, called a MultiSystem. This AS/400 MultiSystem contained over 20 gigabytes of random access memory, 60 processors, and more than 3 terabytes of RAID5 disk. The output of this query test was a 58 million record answer, simulating a retailer's analysis of purchase history over several years period.

This enormous query result was part of a battery of tests performed by IBM to simulate a very large customer data warehouse using DB2/400. This scalability test suite also demonstrated a single AS/400 12-way system performing table scans of over 2 billion records in under an hour. The tests also included a complex "multi-table join" processing millions of records in an hour and a half using all 60 processors in the MultiSystem, and addressed the ability to manage this large environment by demonstrating both backup and restore to tape of 1.1 terabytes of data using IBM's 3590 tape units.

"The AS/400 evolution began in the mid-range where it excelled, then moved into the low-end server market," said Daniel Graham, Global Strategy and Operations Executive, IBM Global Business Intelligence Solutions. "Today we have evidence that the AS/400 is a powerful large scale server. The AS/400 is now a completely scalable server family from the deskside to the multi-terabyte range.

"Based on tests that can be completed only at the Teraplex Centers, AS/400 customers now know there is plenty of headroom in the server family to grow with their business. The AS/400 can scale with this growth. Customers will never have to abandon their investments in skills, software or hardware because of some arbitrary limit in the technology," Graham added.

Built by IBM, the AS/400 Teraplex Integration Center, and two others in Poughkeepsie, N.Y., for the S/390 and RS/6000, are intended to help businesses push the limits of their current data warehousing hardware and software, and design future computing configurations. The three Teraplex Integration Centers, created by IBM at a cost of $47 million, have a combined storage capacity of 12 terabytes, representing one of the world's largest, most powerful dedicated integration facilities. The Centers, which were a focal point of IBM's company-wide business intelligence initiative that was unveiled February 26, affirm IBM's commitment to developing scalable systems, including servers, database software, data mining software, storage and database tools, to support the growing number of business intelligence users around the world.

Business Intelligence involves the gathering, management, and analysis of data for the purpose of turning that data into useful information which is then distributed throughout an enterprise and used to improve decision making. Companies can then make more strategic decisions about which markets to enter, how to select and manage key customer relationships, and how to select and effectively promote products, all in an effort to increase profitability and market share. These practices can also reduce operating costs through more effective financial analysis, risk management, fraud management, distribution and logistics management, and sales analysis.

The Teraplex Integration Centers are open to IBM's customers and business partners for no charge and are equipped for testing of applications and products from multiple IBM divisions and from non-IBM vendors.

More information about business intelligence and the Teraplex Integration Centers can be found at http://www.ibm.com/bi