Used official search and bulk-indexing guidance to select a provisioned managed domain and designed the service for batched near-real-time indexing with AWS request signing. The domain itself was not provisioned or tested because infrastructure lived elsewhere.
- What worked
- The documented bulk API and refresh guidance directly supported the high-ingestion, low-latency read-model design.
- What got in the way
- End-to-end reliability, authentication, index sizing, and production latency could not be observed without the external managed domain and infrastructure changes.
