Ponderings, insights and industry updates

Taxi racing through city at night.
Photo by Jonathan Riley on Unsplash

Hydrolix 18x Faster than Elasticsearch

Published: February 4, 2021

Updated: February 15, 2022

Author: Martin KaganCo-founder/CEO at Hydrolix

Tags: ,

Mark Litwintschik, consultant, tech author and prolific benchmarker of databases, ran Hydrolix through its paces recently and published the results on his blog. Starting with a data set of 1.1 billion NYC taxi rides, Mark walks you through set up, ingest, and querying of Hydrolix on AWS. His results, as shown below, demonstrate Hydrolix’s remarkable performance and economy — for instance, Hydrolix, running on a single c5n.9xlarge query head instance and a single peer, delivered 18x better query performance than a highly tuned Elastic search instance.

Excerpt from Marksblogg Benchmark Summary:

0.4661.0940.7421.412Hydrolix & c5n.9xlarge cluster
8.118.18n/an/aElasticsearch (heavily tuned)
Hydrolix proves 18x faster than Elasticsearch.

Aside from demonstrating Hydrolix’s outstanding performance ingesting and querying time series data, Mark’s benchmark (and his description of our architecture) showcased what makes Hydrolix so groundbreaking: 

  1. a decoupled architecture that allows you to scale ingest and query resources separately, managing performance and costs on a per-job basis
  2. a patented compression scheme that uses S3 for storage, removing the “hot” and “cold” storage distinction (and cost/performance penalties)
  3. on-prem deployment, or the fact Hydrolix runs in your own VPC, which means better security, more control, and zero-egress charges.

With step-by-step instructions on how to install and run the benchmark yourself, Mark also underscores how simple it is to start using Hydrolix. This is a great blog for anyone interested in reducing data storage costs and improving performance for their append-only data workloads and well worth the read. 

— Marty

Share Now