Scaling inside with Swarm64 and Intel FPGA at Strata NYC

Posted in How to
FPGA Postgres reference architecture

Greetings from Gotham!

Swarm64 is at the Intel booth (#717) this week at the Strata Data Conference 2019 in New York City.

We will be showing our newest Swarm64 database acceleration demo, which processes billions of records in real time on a very affordable server equipped with all sorts of goodies from Intel.

Billions of trading records queried in real-time on a single, FPGA-equipped server

Swarm64 is FPGA-based software that accelerates database data loading and query performance. In our new demo, Swarm64 is used to query IEX DEEP order book data as it streams into standard, open source PostgreSQL.

The demo streams millions of rows of data into PostgreSQL every few seconds, ingesting 1 day of trading data in 15 minutes.

While the data is streaming in, Swarm64 is also enabling PostgreSQL to execute SQL statements that perform a standard technical analysis of the share price (50 day rolling average crossing 200 day rolling average).

Then alerts are provided to a human user to analyse a specific stock or traded instrument in more detail as technical indicators have changed. Alerts are, as part of the demo set-up, also texted to a mobile device.

All of this happens on a single server equipped with:

  • Intel® Xeon® C620 Series CPU
  • Intel® Programmable Acceleration Card with Intel Arria® 10 GX FPGA
  • Intel® Optane™ DC
  • SSD storage
  • PostgreSQL 11
  • Swarm64 Data Accelerator software

The amazing thing about the demo is the scale and performance that can be achieved with vanilla open source DBMS and off-the-shelf hardware…no expensive analytic database licenses required and no large clusters to administer.

Scaling inside with Swarm64 database acceleration software and Intel hardware is the easiest path to new insights.

See you at booth 717! Come visit with us to discuss data projects like Netezza replacement, data warehouse modernization, IoT, and others.