Why did Magnite use a GIN index for full-text search instead of a separate search engine (Elasticsearch)?

Using GIN index in PostgreSQL for full-text search instead of a separate search engine like Elasticsearch can be motivated by several reasons:

• Architecture simplicity: there is no need to maintain a separate Elasticsearch cluster, which reduces complexity and infrastructure costs.
• Sufficient performance: for tasks with moderate load and data volume, full-text search based on GIN index can be fast enough.
• Data integrity: search occurs directly in the database, which eliminates synchronization issues between the database and the search engine.
• Resource savings: there is no need to allocate separate servers and resources for Elasticsearch.

However, for large data volumes and high search requirements (e.g., complex queries, scalability, analytics), specialized engines are usually chosen.

Thus, in Magnite, the choice of GIN index could have been driven by a balance between performance, support complexity, and search requirements.