Data
Elasticsearch
Search and log analytics at scale.
Layer
Data
Used in
All services
Same layer
6 technologies
How we use Elasticsearch
Elasticsearch powers full-text search and log exploration when Postgres is not enough.
Where we reach for it
Product and content search
Full-text search with typo tolerance, filters, facets and ranking for shops, catalogues and document libraries.
Log exploration
Centralised logs from several services, searchable by time, field and error pattern.
Multilingual search
Language-aware analysis for German, French, Italian and English content in one index design.
Search beyond the database
Replacing slow LIKE queries when relevance, speed and filter combinations outgrow what PostgreSQL full-text search handles.
When we pick something else
For modest search on a few thousand records, PostgreSQL full-text search is simpler and avoids a second system to run. If a hosted search service fits the budget and data location, Algolia or Typesense can save operations work.
What we watch for
It is not the source of truth
Keep the primary data in your database and treat the index as rebuildable. We plan reindexing from the start.
Mappings and shards
Field mappings are hard to change later, and too many small shards waste memory. We design both before loading data.
Resource use and upgrades
Elasticsearch is memory-hungry and needs monitored heap, disk watermarks and a plan for major version upgrades.
Common questions
- Do we need Elasticsearch or is database search enough?
If search is a minor feature, database full-text search is often enough. Elasticsearch pays off with large data, rich filtering, relevance tuning or log analytics.
- Elasticsearch or OpenSearch?
They share an origin and similar APIs. We choose based on licensing, hosting options and the features you need, and keep the application code behind a thin search layer.
- Can you add search to our existing application?
Yes. We sync data from your database into an index, build the search endpoint and UI, and set up reindexing so the index can be rebuilt at any time.
More in Data
Next step
What is costing you time, money or customers today?
You get an honest assessment of whether and how we can help, followed by a written plan with a cost estimate and timeline.
Tell us what you're buildingWe reply within 1–3 working days, by email, call or video. Your choice.