Query Planner Early Benchmarks

The development team at SpiceDB has been working on a new Query Planner aimed at enhancing the performance of permission checks in the SpiceDB system, which uses a permission graph to determine relationships between resources and subjects. The Query Planner utilizes advisors to optimize path selection and employs dispatching to offload subproblems to other nodes in a SpiceDB cluster, trading network cost for caching benefits. Benchmarks were conducted across various scenarios, such as DeepArrow and WideArrow, to test performance improvements in both algorithmic and datastore contexts. The results indicate that the Query Planner, particularly when advised, generally outperforms the current system in terms of speed and memory efficiency. Although early versions of the dispatch mechanism may be overly active, the team plans to refine this in future updates, focusing on when to send subproblems over the network versus processing them locally.

No tracked trend matches for this post yet.

Use this post, company, and trend context to find content marketing opportunities, perform competitive analysis, or address product feature gaps via the Plushcap MCP server or the Plushcap API.