OpenSearch Deep Dive - Search (part 3)

Michael resumes his deep dive series into the OpenSearch search path after a holiday break, focusing on the ‘TransportSearchAction’ class and the ‘searchAsyncAction’ method. He explains the process of handling search requests, discussing various phases like can-match pre-filter and query-then-fetch. Michael describes how shards are managed, how results are collated, and touches on optimizations to minimize network traffic. He also covers converting coordinator level requests to shard-level requests and highlights some internal workings and challenges in routing and aggregating results. The discussion concludes with dispatching requests from the coordinator to data nodes. The data node query logic will be covered next time.

Part 1: https://www.youtube.com/watch?v=MJPqYT7QPXA
Part 2: https://www.youtube.com/watch?v=TPIa64lJacc

Видео OpenSearch Deep Dive - Search (part 3) канала OpenSearch