You are viewing an older version of this section. View current production version.
In a MemSQL cluster, an
aggregator is a node that functions as a cluster-aware query router. An aggregator stores only metadata and acts as a gateway into the distributed system. A MemSQL cluster may have one or many aggregators depending on query volume.
An aggregator is responsible for:
- querying the leaves
- aggregating the results
- returning the results to the client
The Master Aggregator is a specialized aggregator responsible for cluster monitoring and failover.
Starting a Master Aggregator
A master aggregator node is started as a MemSQL node without additional parameters. Once started, you can run BOOTSTRAP AGGREGATOR on the node to make it a Master Aggregator. The node must not be part of an existing cluster.
Starting a Child Aggregator
A child aggregator node is started as a MemSQL node without additional parameters. Once started, you can run ADD AGGREGATOR on the Master Aggregator to add the node (as an aggregator) to the cluster.