liveBook · Manning

Zero downtime

As explained in Elasticsearch in Action, Second Edition

Zero downtime is continuous service availability during an operational change without application interruption.

Zero downtime keeps an Elasticsearch service available while an alias moves client traffic from an old index to a replacement index. Applications continue querying and indexing through the alias instead of addressing a physical index directly. section 6.2.5

Use index aliases to support an uninterrupted index migration:

  • Create a replacement index with the required settings and mappings. section 6.2.5
  • Point application queries and writes at a stable alias. section 6.2.5
  • Designate an is_write_index when one alias targets multiple indexes. section 6.2.5

Switching a cars index

Keep clients using vintage_cars_alias while migration occurs. Remove that alias from vintage_cars, then add it to vintage_cars_new through one _aliases request. Searches and writes continue through the unchanged alias name.

How Elasticsearch in Action, Second Edition covers it

How deep it goes
Explains alias creation, multi-index write selection, inspection, and coordinated removal and addition, without detailing reindexing procedures.
Examples
Uses cars_for_aliases, vintage_cars, vintage_cars_new, and multi-index car aliases to demonstrate routing and migration.
What you can do after
Operate/configure Elasticsearch aliases to redirect searches and writes during an index migration.
How it connects
Connects index aliases with reindexing, indexing, searching, mappings, and settings.

Also explained in

Compare how 3 books explain zero downtime →

Related concepts

This explainer was generated by Manning's knowledge graph from the text of Elasticsearch in Action, Second Edition. The section links point to the parts of the book it draws on.