Elasticsearch
Elasticsearch is excellent at finding and aggregating documents at scale. It is a poor substitute for your primary database. These notes cover the decisions that keep search useful in production — mappings, writes, queries, capacity, and day-2 operations.
Topics
- When to use Elasticsearch
Search and analytics workloads that belong in ES — and when you should keep the source of truth elsewhere.
- Mappings and analysis
How field types and analyzers decide recall, precision, and whether your index can evolve safely.
- Indexing and ingestion
Bulk indexing, ids, routing, and keeping the search index in sync with the system of record.
- Querying and relevance
Filters vs queries, ranking, and keeping search results useful instead of merely matching.
- Aggregations and analytics
Facets and metrics that stay fast — and how cardinality and nesting turn dashboards into cluster load.
- Shards, replicas, and capacity
How primary shards, replicas, and heap sizing decide latency, resilience, and whether rebalancing helps.
- Operating Elasticsearch
ILM, monitoring, upgrades, and the signals that mean search is about to miss its SLO.