When we originally launched the Varnish API Engine, we sought performance. We’ve continued to develop our API engine with a focus on continuous improvement as we’ve built Varnish API Engine 2.0. We know performance and scalability are underdeveloped areas in the API management space, both in terms of how API gateways will match the current API volumes and how they they’ll cope with the massive increase in traffic ahead. We took the engine out for a spin under several different conditions, and wrote up our test findings to share upon launch of the latest Varnish API Engine.
The document includes the scope, methodology, reference tests, test scenarios and the relevant details about the environment and its configuration. And, of course, the results.
Our testing results indicate that Varnish API Engine delivers a baseline capacity of 23,000 API calls per second on a three-node API engine cluster. We believe that 23K/second is highly achievable in real-world scenarios. Naturally every benchmark comes with the caveat that real-world conditions are very difficult to replicate in a test environment. We prefer erring on the side of caution when estimating performance.
All in all, we’re happy with the performance we got in testing, and we would love to share the details with you. Please take a look at the full testing report to see exactly how we did it while below you can see highlights of the results.