Dell Technologies APEX Block Storage for Public Cloud Multi-AZ Storage Resilience

Krithika Gopal, a Product Manager at Dell Technologies, discusses the deployment and resilience of APEX Block Storage, particularly in a scenario where one of the three availability zones (AZs) is taken offline. The demonstration uses SQL Server 2022 to illustrate the process, but the principles apply to any application.

The APEX Block Storage cluster is set up with six storage instances, two in each AZ. Krithika emphasizes the importance of verifying the system’s health before simulating an outage by shutting down all instances in one AZ. Using PowerFlex Manager, they browse block volumes and check the health statistics and metadata manager to ensure everything is connected and functioning correctly.

The simulation involves manually stopping two instances in one AZ through the AWS portal and observing the impact on the cluster. Despite a temporary dip in transactions per minute (TPM), the cluster remains online, and the application continues to function. The cluster demonstrates self-healing capabilities, as the rebuild process completes in under 30 seconds, restoring the cluster to a healthy state with four nodes.

Next, Krithika restarts the two stopped instances to observe how the cluster rebalances the workload. With all six instances running again, the cluster quickly returns to normal performance after a brief dip during the rebuild. This test confirms the resilience and self-healing nature of the APEX Block Storage cluster in the event of an AZ outage.