STAC-M3 baseline and scale benchmarks of DDN SFA 12K with kdb+ on IBM x3850
Fastest mean response time in the YRHIBID test of all publicly disclosed results using kdb+. Significantly faster than the same tests conducted on the previous-generation DDN storage last year.
26 November 2012 - New York
STAC has released two audited STAC-M3(TM) Benchmarks of Kx Systems kdb+ 2.8 with the Data Direct Networks (DDN) SFA12K-40 Storage Platform and an IBM x3850 X5 server. STAC-M3 provides industry standard benchmarks for tick database stacks.
Results for the baseline STAC-M3 suite (Antuco) are here.
Results for the optional STAC-M3 suite (Kanaga), which tests how well a stack scales with increases in data volumes and concurrent users, are here.
In the baseline tests, this system delivered the fastest mean response time in the YRHIBID test (and corresponding bytes-per-second throughput) of all publicly disclosed results for systems using kdb+, including SUTs using pure solid state storage (benchmarks: STAC-M3.ß1.1T.YRHIBID.LAT2 and STAC-M3.ß1.1T.YRHIBID.BPS). And compared to the DDN-based stack tested last year that used the same server architecture with previous-generation storage (SUT ID KDB110913), this system had mean response times that were 1.2x to 5.5x faster across both test suites (excluding the YRHIBID2 benchmark, which does not use storage). Click here to see all of the reports based on DDN storage.
The landing page for each report contains a link to information and software that is available to employees of firms with premium memberships in the STAC Benchmark Council. This includes detailed configuration information for the stack that was tested, the detailed STAC-M3 Benchmark specifications, the STAC-M3 Test Harness materials, and the kdb+ STAC Pack that was used (so that you can run comparative benchmarks on your own systems).