AlwaysOn Failover Cluster- Explanation

To Start learning about the new SQL Server 2012 feature AlwaysOn lets start with basic study about the Windows Server Failover Clustering.

WSFC is a server feature which is used for the high availability for the critical applications. One such application is SQL Server service which is hosted in WSFC. Moving on to the explanation about AlwaysOn, this is a feature of SQL Server which leverages the Windows Server Failover Clustering(WSFC). The AlwaysOn functionality provides high availability through the redundancy at the server level which is a failover cluster instance. The Failover cluster Instance(FCI) is a SQL Server instance which is installed in the WSFC node and the main advantage of having the AlwaysOn is that these nodes can be geographically dispersed over i.e. it supports AlwaysOn in multi subnets. On a network the FCI appears to be a SQL instance installed on a single node however it provides failover from one WSFC node to another if a current node is unavailable.

A series of events occur when a service is transferred. To describe that a defination of Resource Group is must.To describe resource group, this is the group in which the FCI runs or our SQL instance runs. The resource Group lives in more than one node. So when the FCI starts up, one of the node takes the ownership of the group and brings its SQL Server instance online. The resources in the group lists are-

  1. Network name
  2. IP addresses
  3. SQL Database engine
  4. SQL Server agent

So now lets see the series of events that occur when a failover happens. These are as below-

The dirty pages are written to the disk.
The respective SQL Services in the resource group are stopped on the active node.
The new resource group owner starts its SQL Services.
Client is automatically connected to new node with the help of same virtual network name(VNN).

Please not that in any point in time only one node can run its services.

To keep the FCI online, the WSFC should be in good quorum health. IF the WSFC loses its quorum then the entire cluster is down and manual intervention will be required.

 

Stay tuned for more updates to AlwaysOn.