An Aurora multi-master setup allows you to configure a pair of masters in an active-active read-write configuration which can later be scaled up on demand by the customer to a maximum of four masters. In this configuration, you can read and write to any of the provisioned master instances, providing improved fault tolerance within the compute layer.

In the example shown here, the configuration deploys an active-active pair of compute instances with each instance being deployed in its own availability zone. If an instance outage occurs in one availability zone, all database writes can be redirected to the remaining active instance managed by the customer in the client-side logic, and all without the need to perform a failover. This also provides protection from az outages.

As earlier mentioned, when required, a customer can scale up the multi-master instance count up to a maximum of four instances. Do keep in mind that in this multi-master configuration, you cannot add additional read replicas into the cluster. In terms of connection management, incoming database connections are not load-balanced by the service. Rather, the load balancing connection logic must be implemented by you and performed on the client side. With this in mind, let's now take a quick look at a demo that shows how easy it is to set up and use a multi-master Aurora database cluster, complete with client-side connection management.