OpenSearch · Search and Analytics

Outcome

• Search & AggregationsIAM, Cluster Setup, and the Access Model
  • Explain the three access layers: domain policy → IAM policy → FGAC role mapping
  • Explain why all three 403 failures look identical and the debugging order
  • Map an IAM role to an OpenSearch FGAC role via _plugins/_security/api/rolesmapping
  • Explain when to use VPC vs public endpoint
• Search & AggregationsIndex Design: Mappings and Settings
  • Define an index with explicit mappings: text, keyword, float, date, geo_point, boolean
  • Use multi-field mapping for fields needing both full-text search and aggregation
  • Explain why mapping type changes on existing fields are forbidden
  • Create an index alias and explain the zero-downtime reindex swap pattern
  • Explain what "index": false does and when to use it
• Search & AggregationsIndexing and Bulk Operations
  • Bulk index documents using NDJSON format
  • Explain why Bulk API returns HTTP 200 even when individual documents fail
  • Check .errors and .items[*].index.error in bulk responses
  • Explain near-real-time indexing (1-second default refresh)
• Search & AggregationsSearch: From Basic to Production Queries
  • Use match for text fields and term for keyword fields
  • Build a bool query with must, filter, should clauses
  • Explain why structured conditions belong in filter not must
  • Write a geo_distance filter and sort results by distance
  • Use multi_match with field boosting and fuzziness: AUTO
• Search & AggregationsAggregations: The Search Results Page
  • Build a faceted search returning results + sidebar counts in one request
  • Use terms, range, histogram, and stats aggregations
  • Nest sub-aggregations to compute per-bucket metrics
  • Explain why terms returns 10 buckets by default and set explicit size
  • Explain why aggregations on text fields fail
• Search & AggregationsDynamoDB → OpenSearch Sync Pipeline
  • Build DynamoDB → Streams → Lambda → OpenSearch sync pipeline
  • Handle INSERT, MODIFY, and REMOVE events correctly
  • Explain why forgetting REMOVE creates ghost documents
  • Explain why bulk index is idempotent and makes Lambda retries safe
  • Verify end-to-end: DynamoDB write → searchable in OpenSearch within seconds