STAC Reports: kdb+ tick analytics on two kinds of Samsung SSD

A Z-SSD solution beat or equaled a solution based on 3D XPoint drives in 7 of 17 mean response-time benchmarks

30 May 2018

STAC recently performed the baseline STAC-M3™ Benchmarks on two solution stacks involving Kx Systems kdb+ 3.5 on a Dell PowerEdge R960 server and 4 x Intel Xeon Gold 6154 @ 3.00GHZ processors:

• A solution that utilized 6 x 800GB SZ985 Samsung Z-SSD (KDB180418a). The SZ985 is the first storage device based on Samsung’s Z-NAND technology, designed for latency sensitive, I/O-intensive applications.
• A solution that utilized 6 x 1.6TB Samsung PM1725a SSD (KDB180418b). The PM1725a is an enterprise NVMe SSD that uses Samsung's V-NAND technology.

STAC-M3 is the set of industry standard enterprise tick-analytics benchmarks for database solutions that manage large time series of market data (tick data). In all, the STAC-M3 specifications deliver dozens of test results, which are presented through a variety of tables and visualizations in the reports. Samsung chose to highlight the following:

• In 7 of 17 mean response-time benchmarks, the Z-SSD solution performed better than or the same as a solution with 6 x 750GB 3D XPoint drives from a competitor (SUT ID KDB171010). Including:
  • 1.5x the performance in the single-user interval-statistics benchmark (STAC-M3.β1.1T.STATS-UI.TIME)
  • 1.4x the performance in the ten-user interval-statistics and aggregated statistics benchmarks (STAC-M3.β1.10T.STATS-UI.TIME and STAC-M3.β1.10T.STATS-AGG.TIME)
• In 13 of 17 mean response-time benchmarks, the Z-SSD solution performed better than or the same as the PM1725a solution, including:
  • 6.5x the performance in the market snap benchmark (STAC-M3.β1.10T.MKTSNAP.TIME)
  • 1.8x the performance in the theoretical P&L benchmark (STAC-M3.β1.10T.THEOPL.TIME)
• In 9 of 17 mean response-time benchmarks, the PM1725a solution performed better than or the same as a solution with 6 x 1.6TB 3D NAND flash drives from a competitor (SUT ID KDB170703). Including:
  • 1.8x the performance in the month- and week-high bid benchmarks (STAC-M3.β1.1T.MOHIBID.TIME and STAC-M3.β1.1T.WKHIBID.TIME)
  • 1.7x the performance in the 100-user 12-day VWAB benchmark (STAC-M3.β1.100T.VWAB-12D-NO.TIME)
• In 12 of 17 benchmarks, the PM1725a solution had lower max response times than SUT ID KDB170703, including:
  • Over 45% lower in STAC-M3.β1.1T.MOHIBID.TIME and STAC-M3.β1.1T.WKHIBID.TIME
  • 42% lower in STAC-M3.β1.100T.VWAB-12D-NO.TIME

Details are in the STAC Reports at the links above. Premium subscribers also have access to the code used in this project and the micro-detailed configuration information for the solution. (To learn about subscription options, please contact us.)