Highlight faced performance issues with complex queries in Postgres, prompting a switch to OpenSearch for better scalability and query flexibility. Initial tests showed substantial improvements in query performance, reducing execution times from seconds to milliseconds, even with additional filters. To optimize indexing for frequently updated large objects, they adopted the join data type, modeling errors as parent documents and instances as children, which improved resource usage despite a slight increase in query time. Additionally, they used compound fields to efficiently combine key-value pairs, enhancing query performance without significant impact. A discovery regarding index refresh rates led to adjustments, as OpenSearch's default settings deferred index refreshes if shards weren't queried within 30 seconds, affecting performance during low traffic periods. The adjustments aimed to maintain optimal performance across varying traffic levels.