Operating an OpenSearch Powered E-commerce Search Engine at Scale
summary
Operating an OpenSearch e‑commerce stack for 1000+ stores uses Adelean’s A2 platform with concrete ops rules: autocomplete uses at least four small indices and explicit mappings, function_score is fine if kept clean, flatten nested data for ~10x faster queries (joins can be ~100x slower), keep cluster state small by moving synonyms to files or a dedicated synonym index/plugin, run nightly reindex to remove deleted docs, prefer collapse+cardinality over inner_hits, start with one shard, replicate shards to all nodes, and monitor with Grafana/Prometheus and the profile API.
AI-written summary · about
chapters (15)
- 0:00OpenSearch in e-commerce intro
- 2:31Autocomplete optimization
- 8:20Cluster state & synonyms
- 12:15Index vs query time costs
- 14:02Nested data performance
- 21:23Unifying data models
- 23:10Deleted docs & reindexing
- 26:21Cluster size limits
- 30:42Beyond default limits
- 31:03Replicating indices
- 32:29Debugging search queries
- 34:18Monitoring e-commerce search
- 35:00Vector search use cases
- 38:15Resources & closing
- 39:00Q&A on synonyms & queries
related talks
39mOperating OpenSearch at Scale With Kubernetes Operators: Lessons Learned and Community Insights
28mOperational analytics with AWS
38mMigrating from Self-Managed ElasticSearch to OpenSearch
21mNoOps for OpenSearch: The Future of Self-Healing Search Infras...
24mClosing Remarks & Mastering Efficiency: Pathbreaking OpenSearch Index Management (Ingestion, Index)
32mUnleash Your Cluster's Potential with OpenSearch Benchmark
41mScaling Welfare With Open Source: How NAV Migrated To Managed OpenSearch
45mElasticsearch at OLX: Design, Operation and Maintenance of High-Performance Clusters