What's evolving in Elasticsearch and Lucene?
summary
Elasticsearch 7 and Lucene 8 deliver concrete changes: a “magic wand” precompute of per-term max scores boosts single-term throughput to 1,160 qps, VKD trees replace old numeric indexes improving geo-shape indexing (users report up to 35x faster indexing and 50% faster queries), soft-deletes enable full-history cross-data-center replication and a changes API (soft-deletes), frozen indices trade CPU for disk to hold more cold data while throttling search, and Zen2 removes fragile initial-master settings.
AI-written summary · about
chapters (16)
- 0:00Introduction to Elastic
- 1:11Lucene performance updates
- 2:51Magic Wand optimization
- 5:43Improved Geo-shape handling
- 7:56Changes API and soft deletes
- 9:27Cluster coordination update
- 11:42Adaptive replica selection
- 13:02Cross-cluster search
- 14:33Cross-cluster replication
- 15:35Frozen indices for storage
- 17:27Index lifecycle management
- 18:43Nanosecond timestamps
- 19:10Search as you type features
- 21:08Interval queries
- 23:14Q&A session
- 31:09Advanced performance questions
related talks
37mAlgorithms & data-structures that power Lucene & ElasticSearch
30mCaching for Speed: Practical Performance Optimization in OpenSearch & Lucene
25mApache Lucene: From Text Indexing to Artificial Intelligence
45mAnything New in Search? What If I Find It?
41mTips and Tricks to Scale Elasticsearch for 1M RPMs and Beyond
01h 04mApache Lucene 8, Apache Solr and Elasticsearch – an Overview
37mBeyond the basics with Elasticsearch
55mFull-Text Search Demystified