OpenSearchCon India 2025

Cost for running OpenSearch also always been a huge factor. In spite of OpenSearch being completely open-source and no licensing cost, when you are dealing with terabytes and petabytes of data, cost can quickly stack.

This talk aims to empower OpenSearch users by taking a holistic view using best practices and with practical steps & checklist for reducing the infrastructure cost and to do more with less with OpenSearch regardless of the workload (be it Search/Log Analytics/Observability/Semantic Search).

We'll discuss

1/ Gaining optimal performance from existing resources
2/ How to right size a domain & when to Horizontal scale vs. Vertical scale
3/ How to scale down without impacting performance
4/ How much heap memory is too much or too little and what is the correct vCPU ratio for the given heap memory
5/ When to de-compose larger clusters into smaller clusters Vs. When to aggregate smaller clusters to few larger ones
6/ How to write cache efficient queries, optimize mapping and the index settings
7/ Knobs to improve ingestion and search speed
8/ Budget friendly ways to implement Semantic Search