I'm currently working on a project to consolidate SQL VMs (for 3rd party apps) to a two node VM failover cluster (Active/Passive). We are thinking about implementing a multiple instance clustering, each SQL instance will have its own cluster group and resources (SQL, storage, MSDTC, etc). This architecture has its own advantages and disadvantages as you know. For example: Advantages: one failed cluster group will not impact other applicaitons, vendors can have their own instance (elevated permissions if needed), specific SQL patch can be apply for those applications that requires it, etc. Disadvantages: multiple cluster groups to manage, will need to set up/install a new cluster group and resources (SQL, storage, MSDTC) for new apps, may be too much/complex for what we need. I'm trying to make a case to see if we can consolidate all instances into just one instance. The advantages of implementing a multiple instance clustering in my eyes are just not good enough, especially in our environment we do not allow Vendors access to our SQL Servers and applications are only given the access that they really need (most access is dbowner), and applications are not SQL patch dependent. We have a maintenance window for OS and SQL patching so we can patch the passive node, move the single instance, patch the other node, and then move the instance back. What are your thoughts on this? Do I need more evidence for the one instance approach? Are there other things that I haven’t thought about? Any feedback would be appreciated. Thanks.
- Khoua