Our months-long investigation into slow search speeds led to a solution that not only improved performance but also set the foundation for future scal

The search for speed in Figma

submited by

Style Pass

2024-10-17 15:00:02

Our months-long investigation into slow search speeds led to a solution that not only improved performance but also set the foundation for future scale.

Earlier this year, we set out to improve the way that search works at Figma. While search infrastructure has been a part of Figma since day one, our growth has made it increasingly challenging for our search system to reliably find the content that users are looking for. We wanted to establish a solid search foundation that we can build on in the coming months and years.

Until late 2023, we relied on an older version of ElasticSearch to power search in Figma. Then, we started upgrading to OpenSearch, running as part of AWS’s managed OpenSearch service. OpenSearch is a fork of ElasticSearch, created when the ElasticSearch license changed in 2021. While the two are mostly compatible, small differences have accumulated over the last three years, making the migration more challenging than expected.

I joined Figma’s search team with very little production experience with either ElasticSearch or OpenSearch, but extensive experience in scaling large web services. As always when tackling issues with web services, I set out to understand what was happening under the hood with OpenSearch.