Ascential Software, the leader in information asset management (IAM) solutions, has announced that its DataStage XE data integration solution has completed a one Terabyte (trillion bytes) bulk load benchmark in 1,4 hours on a 64CPU system - more than 500% faster than competitors` published benchmark results - working under the same parameters of prior benchmarks announced by competitors.
"These benchmark results firmly establish DataStage XE as the industry`s performance leader," says Julian Field, GM Ascential Software South Africa. "They also demonstrate the near linear scalability of DataStage`s parallel processing technology which we obtained through our acquisition of Torrent Systems last November.
"Performance has become a critical issue as customers are faced with growing data volumes and shrinking time frames for processing business critical data," adds Field.
In addition, Ascential has taken several important steps to create a new industry standard benchmark that better represents real-world customer data integration challenges. This real-world benchmark - the Enterprise Data Integration Benchmark - will measure how fast data can be extracted from multiple disparate sources and transformed into a common format before being loaded into end-user applications. The benchmark is designed to simulate the data integration requirements of mission-critical applications, such as enterprise data warehousing, business intelligence, analytics, customer relationship management, supply chain management and business performance management.
While previously announced published benchmarks measured only data loading or data extraction from a single source, the new benchmark proposed by Ascential represents the complete data integration process - extraction, transformation and loading. "This reflects today`s real-world usage," says Field.
In the Enterprise Data Integration Benchmark, Ascential`s DataStage XE with the Parallel Extender option extracted, transformed and loaded a full Terabyte of complex data in eight hours, 43 minutes. The Parallel Extender option fuelled DataStage`s performance because it employs intelligent partitioning of the data, which takes full advantage of the parallel processing capabilities of the hardware platform, storage devices and relational database.
The new Enterprise Data Integration Benchmark was run on an IBM P-680 24 processor server and the AIX operating system loading a DB2 relational database, extracting from 192 files spread across 64 disks, transforming 68% of the columns, and integrating 2,6 billion rows by 97 columns of data that were 534 bytes wide. When run on a 64CPU system, DataStage could expect to complete this same 1TB benchmark in just over three hours, and show performance of more than 300GB an hour.
"We`re happy to see Ascential taking the lead on benchmarking what customers actually have to do," says Jane Griffin, Partner, Business Consulting, of Andersen. "Our clients` need for realtime analytics is demanding the movement of hundreds of Gigabytes of data an hour from back-office/front-office systems. As those volumes increase, it`s encouraging to see DataStage ready and able to support such customers with power and speed to accommodate future growth."
"Our research indicates that with data integration volumes increasing and batch windows shrinking, enterprises either must sacrifice analytic capabilities or accelerate ETL (extraction, transformation, load) performance," says Doug Laney, VP, Application Delivery Strategies, META Group. "The ability to parallelise ETL job streams can make the difference between common versus competitive data integration. Furthermore, as ETL performance becomes a high priority for more enterprises, we applaud efforts to establish benchmarking methods that are not OLTP-oriented and don`t rely on extrapolation."
Ascential continues its commitment to setting new standards as an active member of the Transaction Processing Council (TPC), which is helping to define new benchmark specifications and standards for the global IT industry. Ascential is aggressively completing its real-world Enterprise Data Integration Benchmark specifications, which are expected by the second quarter of this year.
* Further details about Ascential`s Enterprise Data Integration Benchmark results are available at http://www.ascentialsoftware.com/etlbenchmark/.
Share
Editorial contacts