An implementation-level description of the architecture that allows Hasura v3 to achieve better-than-Dataloader performance across multiple different data sources.

How do we do it? The user sends us a GraphQL request, which we extract, enrich with relations, permissions, and predicates, then compile before sending to the MongoDB data source.

This post will benchmark Hasura against a handwritten Node.js GraphQL server. The end goal is to show how Hasura performs similarly despite being significantly easier to develop.

While benchmarking Hasura, we noticed Hasura performed better when there was more load on the system. In this blog post we cover basics of processor power management & talk about how we see 50% improvement in latencies by tweaking power management settings.

While benchmarking some tweaks to Hasura we noticed some confusing and misleading/inconsistent results which led us down a bit of a rabbit hole. This is a summary of what we've learned so far, what is left to try, and what we've tried which didn't seem to have much effect.

Leveraging Postgres’ execution plan information to find performance bottlenecks