Searches that used to return fast now crawl, aggregations spike CPU, or data nodes are constantly hot. This is Elasticsearch performance tuning — profiling the actual queries and cluster behavior, not generic checklists.
Tune Elasticsearch for your specific query and data patterns.
Tuned Elasticsearch search performance for Niftycent (niftycent.com) — a US price intelligence engine. Profiled slow aggregations using the profile API, found fielddata pressure from a misconfigured text field, and restructured the mapping to use keyword + search_as_you_type.
You share cluster stats, hot threads, slow search log and profile API output for a few slow queries. I identify the cause and recommend tuning steps. Typically one focused day.
Profiling and a written set of tuning recommendations. Implementation is quoted separately.
Usually expensive aggregations, poorly mapped text fields, fielddata usage, or too many shards per node. The profile API tells us exactly which one it is for your queries.
Some fixes need a new index (mapping changes), most don't (query, aggregation, cache and refresh tuning). I list which is which so you can choose.
Other problems I help with.
Optimize Elasticsearch queries, mappings, indices and cluster configuration.
Details →Implement Elasticsearch for search, analytics and indexing.
Details →Migrate to or upgrade Elasticsearch with minimal downtime.
Details →Find and prioritize the real bottlenecks across your full backend stack.
Details →Relevant production work.
Crawler monitoring millions of products. Processing 30M products daily and comparing them.
Built and maintain the full crawling pipeline — from scraping through Elasticsearch indexing to price comparison.
Price intelligence engine for the US market. Processing large-scale product data.
Built the data pipeline and crawling infrastructure for large-scale US product monitoring.
Describe the problem in a few lines — I'll look at it and tell you what I think.