Vexata: Record-Breaking STAC-A3 Benchmarks for Performance on Financial Trading Systems

Vexata, Inc. announced a record in performance with the completion of the STAC-A3 benchmarks, which evaluate the performance of financial services infrastructure technology.

The company, working in conjunction with Levyx, Inc., a provider of real-time, persistent computing, submitted a combined solution for testing by STAC, the Securities Technology Analysis Center LLC, to illustrate how to deliver performance for back testing of trading algorithms.

With Apache Spark accelerated by Levyx Xenon and utilizing six Xeon SP-based servers connected via FC to the Vexata VX-100F Scalable NVMe flash array, the combined solution set new STAC-A3 records. In one benchmark, the Levyx-Vexata solution achieved 35% more trading simulations per second while using 70% fewer servers.

Complete STAC report (registration required).

Levyx and Vexata estimate that the cost of the new solution is just a third of the previous record holder, providing nearly a 300% improvement in price/performance.

“While financial trading platforms have accelerated compute through next-generation FPGAs and GPUs, system architects have since discovered that, in order to achieve optimal performance, they also need to be able to efficiently move large amounts of data to these computing platforms,” said Reza Sadri, CEO, Levyx. “Through our collaboration, Vexata not only provides a very fast storage layer for accessing the data, but it is so fast that, combined with Levyx’s persistent dataframe layer they deliver an ideal alternative to memory for hosting live datasets in a bulletproof enterprise-grade system. The results that we’ve achieved with Vexata demonstrate what’s possible when you accelerate both layers.“

The company’s scalable NVMe flash arrays enable performance at economics for database and analytics workloads, as well as for emerging AI and machine learning (ML) use cases. Mimicking the approach used by AI and ML frameworks that achieve acceleration using GPUs, the firm’s arrays feature hardware accelerators that enable nearly 20x higher performance than existing all-flash arrays that use traditional storage architectures.

Levyx Xenon platform
Click to enlarge

The Levyx Xenon platform is used in the financial services industry for accelerating back testing and risk-management workloads. Risk analytics acceleration platform speeds up the validation of trading algorithms and strategies leading to lower deployment costs and execution times, while improving time to revenues.

Peter Lankford, director, STAC, said: “Across the trading and investment industry, firms are working to turn large amounts of data into profitable strategies through techniques like machine learning. Backtesting potential strategies is a crucial step in that process, and it requires technology that can handle big data at high speed. Quant funds and banks in the STAC Benchmark Council specified the benchmarks in STAC-A3 as an independent way to measure the ability of software and hardware innovations to accelerate backtesting workloads. The fact that Vexata and Levyx submitted a solution for rigorous STAC-A3 testing shows that they are focused on helping financial firms get new strategies to market more quickly.“

Zahid Hussain, CEO, Vexata, added: “Many hedge funds and the world’s largest financial institutions are trusting Vexata to support their accelerated compute frameworks for AI and ML, built using GPUs and FPGAs. When microseconds matter, Levyx’s Xenon accelerated computing platform running on Vexata’s accelerated enterprise storage offers a compelling high-performance solution for back testing and risk analytics.“

^(*) Performance in the STAC-A3.ß1.SWEEP.SPEED2.BEST benchmark compared to the previous record holder (SUT ID: HPAT171029).

Read also:
Vexata VX-OS 3.5 for All-NVMe Flash Storage Performance
435TB of usable capacity in 6U, NVMe-oF for 40GBE and 32Gb FC fabric interface
2018.06.27 | Press Release