Case Studies in Performance

Executive Summary

The best way to compare the performance of database products is in a head-to-head test using a real application, preferably one of your own. This is especially true when evaluating Caché's post-relational technology, because "standard" transaction processing benchmarking methodologies assume the restrictive "row and columns" format of a relational database. They cannot accurately predict the performance of real applications, which often use complex data models.

Because of contractual prohibitions imposed by well-known database vendors, there is very little published data from "real world" performance tests using relational databases such as Oracle and Microsoft SQL Server. Companies who perform such benchmark tests on their own typically find that Caché outperforms relational databases by a factor of five or more, even without changing application code and just using SQL. This performance differential often increases when applications are optimized to take advantage of Caché's post-relational technology.