InnoDB ClusterSet prioritizes availability over consistency in order to maximize disaster tolerance. Normal replication lag or network partitions can mean that some or all of the replica clusters are not fully consistent with the primary cluster at the time the primary cluster experiences an issue. In these scenarios, if you trigger an emergency failover, any unreplicated or divergent transactions are at risk of being lost, and can only be recovered and reconciled manually (if they can be accessed at all). There is no guarantee that data will be preserved in the event of an emergency failover.

If you cannot tolerate any loss of transactions or data during a failover, instead of using InnoDB ClusterSet as your solution, consider using a single InnoDB Cluster with the member servers deployed across multiple datacenters. Bear in mind that this solution would have a noticeable impact on write performance, as a stable and low latency network is important for InnoDB Cluster member servers to communicate with each other for consensus on transactions.

InnoDB ClusterSet does not fail over automatically to a replica cluster. Because a loss of transactions is possible, and data consistency is not guaranteed, an administrator must make and implement the decision to carry out an emergency failover. If the original primary cluster remains online, it should be shut down as soon as it can be contacted.

InnoDB ClusterSet only supports asynchronous replication, and cannot use semisynchronous replication.

InnoDB ClusterSet only supports single-primary mode for the primary and replica InnoDB Cluster instances. Multi-primary mode is not supported.

An InnoDB ClusterSet deployment can only contain a single read-write primary cluster. All replica clusters are read-only. An active-active setup, with multiple primary clusters, is not permitted, because data consistency is not guaranteed in the event that a cluster fails.

An InnoDB Cluster can participate in only one InnoDB ClusterSet deployment. Each InnoDB ClusterSet deployment therefore provides an availability and disaster recovery solution for a single InnoDB Cluster.

An existing InnoDB Cluster cannot be used as a replica cluster in an InnoDB ClusterSet deployment. A replica cluster must be started from a single server instance, as a new InnoDB Cluster. It is possible to use server instances that are part of a Group Replication group as a replica cluster, but you must completely end the Group Replication group first, and be careful of any customized Group Replication settings that might affect InnoDB ClusterSet. For more information, see Section 8.1, “InnoDB ClusterSet Requirements”.

InnoDB ClusterSet does not support the use of instances running MySQL Server 5.7. An InnoDB Cluster that includes MySQL 5.7 instances cannot be part of an InnoDB ClusterSet deployment.