With Azure Arc, you can run Azure data services anywhere: on-premises, multi-cloud, and edge environments. Azure Arc enabled data services simplify data management with features such as automated provisioning, elastic scaling, and unified management from Azure. Today, we are happy to continue down this journey and announce the preview of Always On Availability Groups for Azure Arc enabled Managed Instance for improved high availability.
Business continuity is a key requirement for many applications. On top of the high availability capabilities of Kubernetes which provides built-in health monitoring, failure detection, and automated redeployment mechanisms to maintain service health, Azure SQL managed instances can now be deployed with additional replicas for increased reliability. With a single deployment command, you can easily deploy an Azure SQL managed instance with a system-managed Always On Availability Group backed by multiple replicas.
Azure Arc enabled SQL Managed Instance Always On Availability Groups offer the following capabilities:
- The three-replica availability group is managed internally, including creation of the availability group, and joining replicas to the availability group created
- All databases are automatically added to the availability group, including all user and system databases like master and msdb. This capability provides a single system view across the availability group replicas
- An external endpoint is automatically provisioned for connecting to databases within the availability group for both read-write and read-only connections
- Ability to run read workloads by connecting to the secondary instances
- Automated failovers and instance redeployment in the event of pod or node failure
Deploying a managed instance backed by multiple replicas is easy. The following command deploys the architecture diagram illustrated above in your Kubernetes cluster.
azdata arc sql mi create -n myinstance --replicas 3
Learn how to deploy Azure Arc enabled SQL Managed Instances with high availability here
See Always On Availability Groups in action in a recent talk below