Alexey Ragozin at RIT++ 2012
summary
Presents strategies for combining memory technologies with full-text indexing to support large-scale, low-latency search, arguing that keeping everything in RAM is usually impractical. Describes building Yandex’s first site search, later moving to Lucene-based toolchains that index data on disk while keeping selective hot-data structures in memory, and recommends the sova search product for small deployments (noting it requires nontrivial configuration).
AI-written summary · about
related talks
36mSearch on the fly: how to lighten your Big Data
30mCaching for Speed: Practical Performance Optimization in OpenSearch & Lucene
36mWhen Semantic Search Breaks in your RAG systems: RAM Walls, Silent Failures, and the Architecture Decisions
20mChallenges in Building RAG Systems (and Solutions Using OpenSearch)
48mUnlocking Next-gen Semantic Search With OpenSearch 3.x
42mIntegrating Graph Context With OpenSearch for Enhanced RAG Application Performance
14mOptimizing RAG with Semantic Caching & LLM Memory
26mLeveraging Moore's Law to Optimize Database Performance