DataMirror Claims Top Benchmark for Data Replication

DataMirror last week reported impressive benchmark results for its data replication offering running on the biggest iSeries iron IBM has to offer. DataMirror says its Integration Suite offering was able to replicate more than 2 billion customer records (equaling more than 1 TB of data) in an hour from a 32-way i5 595 functioning as the source to a similarly equipped target machine. DataMirror conducted the tests, which were designed to mimic real-world settings and which included a range of inserts, updates, and deletes, last fall at the IBM Benchmarking Center in Rochester, Minnesota, and IBM verified the results.

The benchmark results are the highest that any data replication product has been able to achieve on an iSeries system, DataMirror claimed in a press release. And it’s not just DataMirror claiming their product is the best–the IBMers at the Rochester labs said DataMirror’s was the fastest result they had ever seen, according to Nigel Stokes, DataMirror’s chief executive officer. “That’s what they told us, and that’s why we put it in the press release,” Stokes said, adding that IBM maintained full editorial control over the announcement and didn’t object to the statement that DataMirror’s replication software obtained “significantly higher throughput” than any other product. For the record, the lone on-the-record comment from an IBMer to make it into DataMirror’s PR came from Dan Frissora of Rochester’s Benchmarking Center, who called the benchmark results “outstanding.”

For DataMirror, the benchmark results represent nearly a ten-fold increase from its previous benchmark result, which was conducted several years ago on smaller, less powerful iSeries servers, according to Jennifer St. Louis, a product marketing manager with DataMirror. The company says its use of change data capture (CDC) technology in the data replication component of Integration Suite–which is also used in its Transformation Server data replication and iCluster high availability offerings–provides performance benefits and reduces risks compared to using other techniques, such as trigger-based or table-scan solutions.

While IBM is (sort of) giving DataMirror top honors for replication speed on the iSeries server, the title of “Fastest Data Replicator in the West” is up for grabs, acknowledges Stokes, who says he’s very pleased with DataMirror’s results. “It’s not like other vendors don’t have the opportunity to go down there, invest the time, and prove their technology runs any faster. But IBM said they haven’t seen any other product utilizing the iSeries to move that much data that fast,” he says. “We’re laying down the gauntlet a bit. We think that other vendors have significant homework to do.”

DataMirror’s Transformation Server offering pumps data to and from all major database platforms, including DB2/400 and DB2 UDB, Oracle, Microsoft SQL Server, and Sybase. The software was most recently refreshed with administrative enhancements last month (see “DataMirror Updates High Availability and Data Replication Software”).