SQLTeam.com | Weblogs | Forums

Cluster question

High Availability
#1

Hi All,
I've been tasked with the migration of a virtual web server, domain controller, and SQL from external hosting to inhouse. I have created the VM environment, restructured the domain, clustered the Web app server and the SQL server (windows cluster only). Can I make the SQL instance cluster aware without having to uninstall/reinstall?

If not, what do I need to look for, and record, before I start the SQL uninstall/reinstall? Any step by step links are always appreciated!!

Much appreciate the help in advance!

#2

If you want a standard failover cluster instance (FCI) then SQL Server must be installed in the WSFC cluster with shared storage allocated to all nodes in the cluster.

The other option would be availability groups - and that will depend on the Edition of SQL and the number of databases. If you have multiple databases that need to failover at the same time then you need Enterprise Edition for the availability group. If you only have a single database and only need a failover solution for that single database - it can be done with SQL Server Standard Edition (using what is called a BAG).

There are many advantages to using Enterprise Edition and availability groups - such as the ability to create read-only secondary's, offloading reads using read-intent, offloading backups, etc.

For an availability group - you do not install SQL Server into the cluster, you used stand-alone instances on each server and storage is not shared. Each instance has its own storage and data is replicated from the primary to the secondary.

#3

Hi Jeff,
Thanks for the response. I had thought about basic availability groups (I'm running standard edition so that's all I get), however my understanding is they don't provide as quick a failover, and need additional scripts and what not, to copy over logs, jobs, etc. Based upon my research, the more complete solution is to make WSFC and FCI. I was just hoping I'd be able to convert the SQL without having to uninstall/reinstall.

My end goal is to provide the quickest and complete failover solution I can, while covering as many failover scenarios as possible with a single solution.

This will be my first attempt at what I'm trying to accomplish, are there any gotcha's I need to keep an eye out for regarding services, accounts, install methods and the like?

Any insight is much appreciated.

#4

AG's generally failover much faster than an FCI solution - since the secondary is already up and running you don't have to wait for SQL Server to start up.

If your are only looking for high availability (HA) then I would recommend the FCI configuration. The failover could be a bit longer - but should not cause any issues or concerns. It is easier to manage because you don't have to worry about accounts/jobs being copied/built on the secondary - or include code to check for primary before executing.

The biggest concern is making sure the cluster and shared storage are set up correctly. If that is all correct - then installing SQL Server into the cluster is fairly easy.

#5

Thanks, much appreciated!