arangodb/Documentation/Books/Manual/Administration/Sharding

!CHAPTER Sharding

Sharding allows to use multiple machines to run a cluster of ArangoDB
instances that together constitute a single database. This enables
you to store much more data, since ArangoDB distributes the data 
automatically to the different servers. In many situations one can 
also reap a benefit in data throughput, again because the load can
be distributed to multiple machines.

In a cluster there are essentially two types of processes: "DBservers"
and "coordinators". The former actually store the data, the latter
expose the database to the outside world. The clients talk to the
coordinators exactly as they would talk to a single ArangoDB instance 
via the REST interface. The coordinators know about the configuration of 
the cluster and automatically forward the incoming requests to the
right DBservers.