eWeek 12/12/2002 - Borland JDataStore Blazes in Benchmark

By: David Intersimone

Abstract: An article in eWeek has highlighted the excellent cost/performance ratio of Borlands Enterprise Server and JDataStore database.

Borland JDataStore Blazes in Benchmark

An article in eWeek has highlighted the excellent cost/performance ratio of Borlands Enterprise Server and JDataStore database. The article describes Borlands results in an industry standard test called SPECjAppServer2001. This benchmark provides a level playing field for the comparison of Java application server performance.

The two important points that emerge from the article and the benchmark results:

  • If youre considering an application server for a current or upcoming project, you can do no better than Borland Enterprise Server.
  • If you need a powerful, economical database for deployment of business critical applications, JDataStore is the top choice.

Background and Analysis

Borland chose to run the benchmark in a single node configuration using a relatively modest systemjust as most customers would. Most benchmark submissions are run on dual or multi node systems with the application server residing on one high-end server and the database on another. Borland ran both Borland Enterprise Server and JDataStore on a single, relatively modest, Dell 4600 dual CPU computer. The results show that Borland delivered far more than 50% of the performance of systems that competitors used that are more than twice as powerfuland considerably more expensive. The result is a very low cost/performance ratio.

Currently, only Oracle, using three high end Hewlett Packard servers in the multi-node category achieved a lower $/BOPS (cost/performance measure). The difference, of course, is that the Borlands Dell system would cost a customer several thousand dollars while the Oracle system would cost many thousands of dollars.

In the single node category, Borlands delivered a result of $500.36/BOPS. IBMs single node configuration delivered a much more expensive result of $ 4847.59/BPOS!

As the article points out, the Borland results mark the first time that a Java database has been used in a published SPEC jAppServer2001 benchmark. To date, all other submissions have been run on extremely expensive, native databases. Borlands high performance in a single node configuration is due in large part to JDataStores ability to provide high performance and scalability for demanding OLTP based applications.

Please read the eWeek article at: "JDataStore Blazes in Benchmark" http://www.eweek.com/article2/0,3959,762562,00.asp


Server Response from: ETNASC03