Speedscale is one of the first commercial technologies utilizing actual API traffic to generate tests and mocks. Speedscale helps Kubernetes engineering teams validate how new code will perform under production-like workload conditions. By leveraging traffic to generate sophisticated mocks, engineers and testers are automatically granted the ability to isolate and performance test smaller components in the context of a tightly coupled architecture. This superpower enables rapid testing iterations. Moreover, without scripting, testing can finally move as fast as development.
Speedscale initially implemented Elasticsearch and exposed the data through Kibana. It allowed them to prototype the use cases quickly and worked great for lower volumes of data. But it scaled poorly, and they had very little control over the look and feel of the UI. They started looking for alternative technology that could run inside Kubernetes, is easy to operate, and is scalable. They evaluated Timescale, InfluxDB, Prometheus, and Graphite.
They selected Timescale because they were already using PostgreSQL as part of their technology stack, and the testing confirmed that Timescale would scale well at their load ranges. Moving to Timescale reduced Speedscale’s AWS cloud costs and dramatically increased query performance.
After implementing Timescale, our AWS cloud costs went down about 35 % because it is cheaper to run than the AWS OpenSearch we used before. In addition, the query performance improved dramatically. A majority of queries take under 100 ms to complete.
Ken Ahrens, Co-founder and CEO of Speedscale