Weka beats DDN, Supermicro Optane and Dell EMC PowerScale in stacks of STAC benchmarks

STAC benchmarks show WekaIO filesystem software running on Amazon cloud EC2 instances can deliver record-breaking tick analytics performance for the financial services industry, beating on-premises DDN EXAScaler, Dell EMC PowerScale and Supermicro Optane systems.

The STAC benchmarks are designed and run specifically for the financial services industry, with audited results by the STAC Benchmark Council. This council consists of over 400 financial institutions and 50 vendor organisations. STAC benchmarks are rigorous, with extensive documentation and none of the potential fudges used in some other storage performance benchmarks. 

Shailesh Manjrekar, Head of AI and Strategic Alliances at WekaIO, said “Capital markets around the world need to analyse more data in less time at the best economics. Public cloud environments were not considered suitable for these workloads due to stringent latency and performance requirements.”

The latest STAC benchmarks show that AWS can be used instead of or augmenting on-premises servers. A supportive quote from Eric Burgener, Research VP, Infrastructure Systems, Platforms and Technologies Group, IDC, said “WekaIO’s latest benchmark results using STAC-M3 show that the vendor can deliver record-breaking performance not only on in-house infrastructure but also in the public cloud using Amazon EC2 NVMe instances.”

Weka STAC history

Weka produced good STAC benchmark results in June 2019, using the M3 baseline Antuco and scaling Kanaga suites for high-speed analytics on time series data running on-premises Penguin servers. It produced more STAC wins in June last year using a setup with 32 HPE Proliant servers.

Now it has repeated its benchmark-topping performance, with the M3 Antuco and Kanaga suites, using AWS cloud compute instances instead of on-premises servers.

Test details

This testing was performed on Amazon EC2 Non-Volatile Memory Express (NVMe) instances using a kdb+ 4.0 database by KX Systems. There were 15 database server nodes and 40 storage nodes.

Weka says its software:

  • Outperformed all publicly disclosed results in three of the five throughput benchmarks in the STAC-M3 Kanaga suite (STAC-M3.β1.1T.{3,4,5}YRHIBID.BPS)
  • Outperformed all publicly disclosed results in three of 24 mean-response-time benchmarks in the STAC-M3 Kanaga suite
  • Was faster in 16 of 24 Kanaga and nine of 17 Antuco benchmarks versus a kdb+ 4.0 solution running on a ten-node cluster with 60TB of Optane persistent memory (KDB200603),
  • Was faster in 20 of 24 Kanaga benchmarks and four of 17 Antuco benchmarks versus a kdb+ 3.6 solution on a DDN EXAScaler parallel file system with 15 database servers accessing all-flash storage appliances (KDB200915),
  • Was faster in 15 of 17 Antuco benchmarks versus a kdb+ 3.6 solution involving nine Dell EMC PowerScale F200 database servers accessing networked flash storage (KDB200914).

Weka says these results show Quants and FSI professionals can run algorithmic trading, quantitative analytics, and back testing use cases in the AWS cloud, getting the low latency they need while taking advantage of AWS elasticity and scalability.

Interested readers can download or view the full KDB210507 STAC report here.