PostgreSQL TPC-H SF1000 benchmark with Swarm64

20x faster PostgreSQL queries (TPC-H – 1TB)

TPC-H is a popular SQL data warehousing benchmark. With 1TB of data, running on a 144-virtual core server, Swarm64 DA improves PostgreSQL 12 query performance as follows:

  • 20x faster overall (all 22 queries)
  • Individual TPC-H queries improve anywhere from 7x to 60x faster
  • With Swarm64 DA, all 22 TPC-H queries execute; without Swarm64 DA, five of the TPC-H queries timeout after 15 minutes.
Swarm64 DA technical architecture

Linear PostgreSQL query speed improvement (1TB TPC-H)

Swarm64 DA extensions enable PostgreSQL query performance to scale linearly as CPU cores are added to the server. Swarm64 DA enables PostgreSQL to exceed its parallel worker limits and apply much greater CPU resources to queries.

As shown in the chart to the left, With Swarm64 DA, increasing the server core count 125%, from 64 to 144, improves PostgreSQL query performance a near-linear 122%. Without Swarm64 DA, queries only speed up 12%.

Webinar: PostgreSQL Parallelism Do’s & Don’ts

PostgreSQL – 5x better user concurrency than MongoDB (customer benchmark)

Wind turbine analytics software company, Turbit Systems, chose Swarm64 DA-accelerated PostgreSQL over MongoDB for their one terabyte time series database.

Swarm64 DA enabled query response times to remain sub-second for 5x more concurrent users than MongoDB. This enabled Turbit to grow their user base 5x without increasing their cloud database infrastructure costs.

Continuously improving Swarm64 DA (TPC-H 1TB)

A long-time Swarm64 DA user benchmarked the performance of Swarm64 DA over several releases of Swarm64 DA.

As the chart shows, Swarm64 DA acceleration of PostgreSQL query performance has tripled from version 3.0 in 2019 to v5.0 in 2020.

We are continuously developing new ways for Swarm64 DA to accelerate PostgreSQL. Give Swarm64 DA a try for free, and give us your feedback. We are always looking for new query patterns to accelerate.