We are using both SQL Server 2012 and 2014 always on.

Here is my issue: we have an application that I have set up to use Availability groups(2014), however the issue I am having is that it uses SSAS and as far as I can tell I cannot set up the SSAS database to be akin to an availability group which floats between two replicas based on the av listener name/IP.

My hope was to have the SSAS database failover to the same replica that the databases would in case of a failover. Is this even possible and would you be able to point me to any articles.

I am referring to the OLAP database that appears under databases when you connect to SSAS through the GUI.


There is nothing like AlwaysOn for SSAS databases.

If you want high availability for your SSAS databases you need to set up a Windows Server failover cluster.

Refer to the documentation for instructions:

One approach for meeting your organization’s high availability requirements for Analysis Services is to install it on a Windows Server failover cluster (WSFC). A WSFC is composed of multiple servers that offer redundancy in the event of a hardware failure, system crash, or some other service disruption. A WSFC can offer protection of your mission critical backend servers, including the Analysis Services instances running your BI workloads.


I realise this is an old question, but you could solve this by scaling out Analysis Services. This technet article from 2005 is still relevant to SSAS 2016.

Basically you process the cube on one server, and synchronize it to the other server. Synchronizing will only copy the changes, making it generally smaller and faster than a backup/restore. You can automate it with a simple XMLA script.

The final part of this is to redirect traffic on a failover. You could do this by changing the connection string in your app (you might be able to change it dynamically) or using a DNS CName (just change where it points to and flush dns). Both require manual intervention. You could look at some sort of load balancing, which could then automate and only send requests to an active server.

The other advantage of this, is you can allow the cube to be queried at the same time as dimensions/measure groups are being processed.

You will need to think about licensing. If you are actively querying both, then both servers will need to be licensed. We have not had a problem with keeping a synchronized cube at our DR site that isn't queried, it was considered to be covered by Software Assurance, but you'll need to speak to a licensing expert to confirm your setup.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.