Breaks the STAC-A3 Performance Record Running on Google Cloud Platform
Levyx Inc., whose high-performance, ultra-low latency data processing software dramatically reduces Big Data infrastructure costs, revealed today record shattering independent testing results performed by the Securities Technology Analysis Center (STAC®) using the STAC-A3 benchmarking suite.
Levyx broke the performance record running the benchmark on four 64-vCPU Google Cloud nodes equipped with SSDs while running Apache Spark and Levyx’s Xenon software. The net result was 32X the performance employing much less equipment in comparison to the baseline STAC-A3 benchmark which used a 14-node (24 cores each) disk-based Map-Reduce Hadoop cluster.
STAC-A3 is the industry’s primary performance benchmark suite for the infrastructure that banks, hedge funds and other trading firms use to backtest potential trading algorithms. Peter Lankford, Director of STAC, said: “Trading firms in the STAC Benchmark Council specified the STAC-A3 benchmarks in order to measure the potential of software and hardware innovations to accelerate backtesting. The competitive pressure on firms to bring smarter algorithms to market in less time, together with the increasing use of machine learning to automate development of candidate algorithms, has put backtesting on the critical path to revenue. Levyx’s use of STAC-A3 shows that the company is putting serious focus on the industry’s desire to speed up this key workload.”
Reza Sadri, CEO of Levyx, said: “Since the financial sector typically pushes the envelope for low-latency technologies, our results for this class of sophisticated backtesting validate our performance benefits in a real-world application and point to an exciting market opportunity within the financial space. Similarly, these performance benefits can also accelerate the Time-To-Value in other large markets, such as analyzing trends in insurance, credit scores and consumer sentiment, to name a few. In addition, the STAC results highlight that we drastically reduce the data center footprint and related costs (in the cloud or on-premise).”
Levyx’s Xenon™ leverages the high-performance nature of Levyx’s Helium™ core engine and extends it into a low latency, scalable data analytics solution. Xenon is designed to manage the retrieval, processing, and indexing of very large datasets, i.e., collections of billions of objects, spread across a tightly coupled cluster of servers, each with multi-terabyte persistent storage capabilities. More specifically, Xenon is a distributed database system having the following functional capabilities:
- • Core SQL functionality: filter, projection, selection, sort, join, groupby, and aggregates on structured data, i.e., schema-based tables.
- • Support for random lookup and neighborhood search using an index rather than scan and filter.
- • Tightly integrate with the Apache Spark system for ease of deployment and use (also fully capable to function in native mode, or serve as an off-load layer for other big-platforms and systems).
- • Scale with the number of cores in the cluster and use SSDs (or other high-bandwidth, low latency persistent storage) as a persistent memory layer for large, live datasets.
The efficiency with which Xenon can process massive workloads can also be applied in the processing of similar large-scale data sets in other applications within the financial sector, as well as in different vertical industries such as Government, Internet of Things (IoT), Oil and Gas, Machine Learning, Artificial Intelligence, and Cybersecurity.
Levyx’s software solutions fundamentally disrupt the economics of Big-Data applications, bringing the benefits of high-speed Big-Data processing to the masses. No longer reserved for the largest enterprises, Levyx technology can process hundreds of millions of queries per second on commodity servers on a few nodes, making Big-Data processing much more accessible to organizations of all sizes.
STAC® is a technology-research firm that facilitates the STAC Benchmark™ Council, an organization of leading financial institutions and technology vendors that specifies standard ways to assess technologies used in the financial markets. The Council is active in an expanding range of low-latency, big-compute, and Big-Data workloads.