The Search Stack That Nearly Broke Us - Until We Did This | HackerNoon
Briefly

Initially, finlight.me leveraged Postgres' full-text search for efficient querying of financial articles, benefiting from its simplicity and performance. However, as the volume of articles surged and user queries became more intricate, limitations in the Postgres architecture surfaced. This led to degraded search performance, particularly in combining free-text queries with additional filters. Ultimately, the move to OpenSearch was necessitated not just by performance issues, but also by the desire to maintain Postgres as the source of truth for data integrity during the migration process.
Keeping Postgres as our source of truth during our transition to OpenSearch proved to be a pivotal decision, allowing us to maintain data integrity.
While Postgres full-text search served us well initially, the system began to struggle under increasing complexity and volume, prompting our shift to OpenSearch.
Our decision to move from Postgres to OpenSearch stemmed from the need for enhanced search capabilities as our article volume grew beyond just a few thousand.
The combination of fast, efficient full-text search in Postgres became insufficient as user queries evolved, revealing the critical need for a more robust solution.
Read at Hackernoon
[
|
]