Reverse search: materializing search results
summary
Describes a monitoring service that processes millions of metrics from agents and focuses on optimizing costly metric-search operations by materializing search results. Uses an approach that avoids simple time-based caching—ensuring newly arrived metrics are immediately visible for graphs and trigger checks—so repeated queries (like trigger evaluations) are accelerated without delaying fresh data. The talk discusses techniques for storing or maintaining search results to balance performance and real-time visibility.
AI-written summary · about
chapters (6)
- 0:00Introduction
- 4:42Storing and indexing metrics
- 10:48Query caching problem
- 17:36Reverse search solution
- 23:00Cache implementation results
- 24:29Q&A
related talks
36mSearch on the fly: how to lighten your Big Data
29mReasoning about performance (in the context of search)
42mIntegrating Graph Context With OpenSearch for Enhanced RAG Application Performance
20mUnified Performance Observability in OpenSearch: Integrating Workload Management and Query Insights
43mThe denormalized query engine design pattern
43mIn Search of Lost Performance
47mWhy, What, When and How of Running Diff Types of Sear...
41mBeyond the basics with Elasticsearch