The Analytics Service provides a parallel data-management capability; allowing the running of complex analytical queries.
The Analytics Service supports large join, set, aggregation, and grouping operations: these activities are anticipated often to employ large amounts of data, and therefore to be highly consumptive of processor, networking, and memory resources; highly demanding in terms of cross-node coordination; and of extensive duration.
Analytic queries can be either predetermined or ad hoc; predetermined queries often providing greater efficiency. The Analytics Service further enhances efficiency by supporting parallel query-processing and bulk data-handling; and by allowing analytic queries to be run on its own form of indexes. These are kept local to the data that has been shadowed to the Analytics Service; and are thus partitioned in the same way that the data itself is partitioned.
Deploying the Analytics Service
The Analytics Service relies on the Data Service, which must therefore be running on at least one of the cluster nodes.
Due to the large scale and duration of operations it is likely to perform, the Analytics Service should be run alone, on its own cluster node, with no other Couchbase Service running on that node.
For the practical steps required to initialize or join a cluster, and to deploy services, see Create a Cluster. For information on how to run analytic queries, see the Introduction to Couchbase Analytics.