Oracle Express Server, the leader in analytic power and flexibility, has once again demonstrated the fastest calculation and query performance of any on line Analytical Processing (OLAP) server on any platform. Oracle Corporations has announced record results for Express Server running the industry standard OLAP Council Analytical Processing Benchmark -1 (APB -1). In a subsequent benchmark, express also demonstrated unprecedented scalability by querying against 50 times more data than Arbor Essbase, in less elapsed time. Achieved on a Unix platform using a Sun Enterprise 450 Workgroup Server running Solaris 2.6, Oracle`s record setting end to end run time was 34 percent faster than results posted earlier this year by Arbor on Windows NT 4.0 for the 0.1 scale factor APB-1 benchmark. Express achieved a Benchmark specified Analytical Queries per Minute (AQM) metric of 8,073-52 percent better than Essbase. The APB-1 benchmark was developed by the OLAP Council in order to reduce confusion in the market over contradictory vendor claims, providing an objective measure by which to judge OLAP server performance. The benchmark has two phases: data load and calculation, and query. The first phase measures the time to load incremental data into an existing OLAP database and perform sophisticated calculations like multidimensional consolidations, cost allocation models and forecasts. The second phase measures the time to process 250,000 OLAP queries. Oracle Express bettered the incremental load and calculation performance of Essbase with a time of 10 minutes 43 seconds, and dramatically outperformed Arbor in the query phase with a time of 20 minutes 15 seconds-an improvement of 44% over Essbase. These results clearly demonstrate the performance advantage Oracle Express offers both administrators-who must perform data loads within a narrow "window" of system availability and users, who demand rapid responses to even the most complex analytical queries. Express Demonstrates Dramatic Scalability While the APB-1 provides a valuable yardstick for evaluating OLAP servers, the relatively small amounts of data used in earlier benchmarks at the 0.1 scale level are not reflective of many real world applications. Oracle is the first vendor to announce audited APB-1 benchmark results with databases larger than 0.1 data density. The dramatic scalability of Express Server was demonstrated by running a 5.0 data density benchmark (i.e., 50 times the input data volume of Arbor`s benchmarks) without modifying a single line of code. Both benchmarks were run on four-processor Sun Enterprise 450 servers. The results indicate that query performance for Express is virtually independent of data volume. Despite the 50-fold increase in data volume, Oracle`s query time of 34 minutes 2 seconds was still faster than Arbor`s performance against a 0.1 scale database. "Oracle Express is widely recognised as the most powerful OLAP server in terms of analytical power and deployment options," said Jeffrey Stamen, senior vice president of the OLAP Products Group at Oracle. "These new APB-1 benchmarks clearly indicate that Express is also the performance and scalability leader." Oracle Express Server delivers enterprise OLAP capabilities, enabling users to analyse data in a structured or ad hoc form using a multidimensional data model that includes support for "what-if" scenarios, forecasting and financial modeling. Express Server is the foundation of the Oracle Express family of OLAP products, including Oracle Express Objects, an object-oriented development environment; Oracle Express Analyser, a general purpose reporting and analysis tool; and Oracle Financial Analyser and Oracle Sales Analyser, pre-built applications for finance and sales/marketing users. Complete Oracle Express APB-1 Results Oracle implemented the APB-1 at 0.1 data density using Oracle Express Server 6.1 hosted on a Sun Enterprise 450 server running Solaris 2.6. The server was equipped with 4 CPUs and 1 gigabyte of RAM. The scalability test was run at a 5.0 data density scale factor (i.e., 50 times more data than the baseline run) on the same server with 4 gigabytes of RAM. Audited results were as follows: 0.1 data density: * Total time for incremental load and OLAP calculations: 10 minutes, 43 seconds * Total time for 250,000 queries: 20 minutes, 15 seconds * Total elapsed time: 30 minutes, 58 seconds 5.0 data density: * Total time for incremental load and OLAP calculations: 259 minutes, 25 seconds * Total time for 250,000 queries: 34 minutes, 2 seconds * Total elapsed time: 293 minutes, 27 seconds Complete, audited APB-1 results and disclosure documentation are available from Oracle`s Web site at www.oracle.com/products/olap/html/bnchmark.html.
Share
Editorial contacts