Installation

Before you try to install a cluster, it is important that you know how a cluster in MySQL Cluster spreads out over multiple computers. It does so by creating three different types of nodes, which do three different jobs:

• Storage nodes These nodes hold the data in the cluster and do most of the actual work processing queries.
• Management nodes These nodes keep the configuration of the cluster and provide a central point from which to manage the cluster.
• SQL nodes These nodes are typically MySQL daemons that accept, parse, and optimize queries and do some of the processing work before returning the results. Some binaries that you will meet later on (such as ndb_restore for restoring backups) are also SQL nodes because they connect to the cluster; however, they have little to do with SQL queries.

You can run SQL nodes on the same physical machine as storage and management nodesand you can run more than one storage node on each physical machine. However, if you want a highly available, fault-tolerant cluster with no single point of failure, you must have at least three physical machines and a management node that does not share a physical machine with a storage node.

The following section discusses a standard installation with the cluster communicating over TCP/IP. Later chapters cover some slightly more complex setups involving high-speed interconnects.