Analytics OpenSearch

Indexing Pipeline: Load ⟶ chunk ⟶ embed ⟶ store

Service = capacity/cluster MGMT

• Serverless = automatic scaling/maintenance. (this one)

  • vector engine provides vector storage and search capability

Amazon OpenSearch Serverless

• is a: managed engine

• Primary purpose: search & vector retrieval

• Stores: docs + vectors + metadata

• Query model: semantic/keyword/vector searches

• Scaling & ops: autoscaling for search workloads.

• Data strength: text + metadata + similarity

Compare to Aurora/pgvector: OpenSearch simplifies scaling & improves retrieval performance.