Skip to content

Xapiand is a fast, simple and modern search and storage server built for the cloud. It features:

  • MIT license (a few files carry their own; see the Copyright Notice).
  • You can find the source code for Xapiand at GitHub
  • Document oriented, in JSON, YAML or MessagePack.
  • No need for upfront schema definition: field types are detected, or declared in a schema.
  • HTTPS with --tls-cert and --tls-key. The API has no authentication.
  • One process: the server, the search engine and the storage, with no separate services.
  • Runs as a single node.
  • Built on top of strata indexes.
  • Full-text queries with stemming and stop words in many languages, fuzzy, partial and wildcard terms.
  • One search can span many indexes, by list, wildcard or _all.
  • Each index also offers storage of files: à la Facebook’s Haystack 1.
  • A stored blob uploads as a stream, past the 16 MiB limit of a buffered body.
  • Metrics aggregations.
  • Bucket aggregations.
  • Lua scripts update documents, with bounds on the instructions, the work and the memory each call uses.
  • Support for more than one index, and indexes nest (twitter and twitter/user).
  • Index level configuration:
    • Schema.
    • Number of shards.
  • Reads see every write the node has answered.
  • Event driven asynchronous architecture using the Asio-based reactor library.
  • Written in modern C++.
  • Prometheus metrics at /:metrics.
  • Uses the Hierarchical Triangular Mesh 2 for indexing.
  • Accepts multiple Coordinate Reference Systems, including WGS84.
  • Implements EWKT.
  • Each index splits into shards on a single node (number_of_shards).
  • A write-ahead log recovers the writes of a crash.

1 Finding a needle in Haystack: Facebook’s photo storage.

2 The Hierarchical Triangular Mesh.


The Xapiand Core Team’s responsibility is to ensure the development and community around the Xapiand ecosystem thrive.