Skip to content

Elasticsearch

Elasticsearch runs a search service over records your app sends to it and can search across text and structured fields.

Visit Elasticsearch

How does Elasticsearch get the records it searches?

Your app sends documents to an index, which is a copy organized for fast searching. Decide which fields are text to search and which fields are values to filter on, such as a ticket category or an account ID. Keep record IDs stable so your updates and deletions reach the right result.

An accepted update might not show in search right away. A refresh makes the latest changes searchable. Check the result after it has refreshed before deciding that an update or deletion failed.

How do you keep one customer’s results out of another customer’s search?

Put the search endpoint behind your app’s server. Check which account is signed in and which records it can read. Keep broad Elasticsearch credentials on the server and return only the allowed fields.

Elastic also offers document and field restrictions for read roles where supported. Check the deployment and subscription before relying on them. A second role with unrestricted access can override those restrictions. Review the roles together and keep writes separate.

How can you plan a private search and a removal check?

Use only paper and fictional records. Draw two accounts with an invoice each. Use the same search word for both invoices so account access determines the result.

  1. Trace each account separately
    Write which invoice each signed-in account should get back for the shared word. An account that cannot read either invoice should receive neither title, snippet nor count.
  2. Follow one deletion
    Cross out the first invoice and mark the next index refresh. After that refresh, the first account should have no hit. The second should still have its original invoice.

What does managed or self-managed Elasticsearch cost you?

Elastic offers a self-managed Basic tier without a license fee. You provide the servers, updates and recovery. Elastic Cloud Hosted bills for resources, and Elasticsearch Serverless meters work such as indexing and querying, stored data and outgoing transfer. Check prices for your deployment and region.

Price the workload you expect, including changes and storage, rather than multiplying a headline rate by the number of searches. Check whether the access features you need are in the plan.

What should you save if you want to move or rebuild search?

Keep the source records, field definitions and the rules your app uses to decide which results people see. Hosted and self-managed deployments can use compatible Elasticsearch snapshots for recovery. Elastic’s Serverless documentation does not offer user-managed snapshot and restore for projects.

Test recovery on a separate environment with fictional data. A snapshot has version constraints, and it does not replace the source data you need for a different search engine.