Description
One MySQL server cannot keep up with the load, replication lags, and a dead node means a manual failover. Surviving failures and scaling out takes a different architecture.
MySQL Cluster is built on the NDB storage engine and uses a shared-nothing design: management nodes coordinate, data nodes store, and SQL and NoSQL clients ask a management node first and then talk to data nodes directly. Writes replicate synchronously, so losing a machine does not take data offline.
Synchronous replication: new data is written to several nodes at once so it stays available.
Auto-sharding: data spreads across data nodes automatically and you add capacity by adding nodes.
SQL and NoSQL access: use MySQL servers for SQL or native APIs straight to the data nodes for low latency.
Central management: one or more management nodes configure and monitor the whole cluster.
MySQL Cluster is built on the NDB storage engine and uses a shared-nothing design: management nodes coordinate, data nodes store, and SQL and NoSQL clients ask a management node first and then talk to data nodes directly. Writes replicate synchronously, so losing a machine does not take data offline.
Features
Synchronous replication: new data is written to several nodes at once so it stays available.
Auto-sharding: data spreads across data nodes automatically and you add capacity by adding nodes.
SQL and NoSQL access: use MySQL servers for SQL or native APIs straight to the data nodes for low latency.
Central management: one or more management nodes configure and monitor the whole cluster.
