June 24, 2016 at 9:29 am
Hello! Can you help me with SQL 2014 SSAS design?
I am planning to install SQL 2014 on two servers and create AlwaysOn Availability Group. I want to install then SSAS on both servers and create Failover Cluster Role for it. The question is - will it work? I found one official paper here, that says it won't (While the AlwaysOn Availability Groups feature also requires a WSFC, it can only be utilized with the relational engine and cannot be used with Analysis Services databases.) and official page here which says it would (If you are using availability groups as your high availability solution, you can use a database in that group as a data source in an Analysis Services tabular or multidimensional solution.).
So the questions are:
1 - Can I use Availabilty Group for hosting SSAS database?
2 - Will Failover Cluster works for SSAS clustering?
One more thing I am thinking about: can I use NLB for balancing SSAS instances? As far as I found NLB support only read activity with SSAS.
June 25, 2016 at 8:03 am
First, are you trying to create HA for SSAS? Meaning you want SSAS to live on the cluster and fail from one node to the other? I did see this, which I think works for FCIs: https://www.mssqltips.com/sqlservertip/3692/clustering-sql-server-analysis-services-part-1/
Or do you want to have SSAS as a separate piece of software on Node 1 that does something, and then have SSAS on Node 2 doing something else? Other cubes?
AFAIK, when you cluster in Windows, there isn't any read access to the passive node. Availability Groups do this, but that's the database level, not a cluster at the instance level.
June 25, 2016 at 8:19 am
I want to cluster SSAS. I want to use availability groups for SSAS database itself and usual WFCL for clustering SSAS role. The first paper I mentioned says that I can't use availability groups, so I want to clarify, what can I do.
I have two cubes and I want one SSAS instance working with them at one moment. The second instance only for failover.
June 25, 2016 at 10:45 am
I think you can do this. It appears the WSFC will work with SSAS, which I think means NLB isn't needed. The resource should appear as a clustered item with a virtual name.
June 25, 2016 at 11:03 am
What about SSAS database? Why first document says it is not possible to use availability groups?
June 25, 2016 at 1:00 pm
Don't confuse clustering (WSFC or FCI, depending on Windows version) with Availability Groups. They're separate.
I'm not an SSAS expert, but there isn't a database per se, correct? Isn't there a model (multidimensional or tabluar) that you connect to, build cubes from, and process data. These aren't databases in the way that the SQL Server database engine has databases.
You can cluster SSAS, but not use it with AGs, AFAIK.
June 27, 2016 at 12:55 am
So I can use WFCL for clustering SSAS instance, and the same WFCL for creating AG for Database enginge, but I can't link them together (however I can use AG database as source for SSAS still), right ?
June 27, 2016 at 8:50 am
I'm not sure. I believe that the underlying WSFC can work for an AG.
As far as using the database in the AG for SSAS, I assume using the virtual instance/db will work fine, but you should certainly test and get someone from Microsoft to note this is supported.
To be clear, is this what you want?
Node A
- node 1 of WSFC, hosting virtual node C
- SQL instance with Database X (and potentiall database y) in Availability Group J
- clustered instance of SSAS, using database X (and /or database Y) on virtual node C
Node B
- node 2 of WSFC hosting virtual node C
- SQL instance with Database X (and potentiall database y) in Availability Group J as failover.
- clustered instance of SSAS, using database X (and /or database Y) on virtual node C
June 27, 2016 at 8:58 am
Yes, that's right. Database X for other purposes and database Y for SSAS.
Viewing 9 posts - 1 through 8 (of 8 total)
You must be logged in to reply to this topic. Login to reply