January 8, 2013 at 4:41 am
Hi all
I'm hoping someone can help me with a question I have regarding mirroring to a cluster. I'm familiar with having a single cluster that contains a database that is mirrored to a remote non-clustered DR server. However...
Suppose we have a production SQL cluster, 2 nodes, active/passive.
Also suppose we also have a DR cluster SQL cluster, 2 nodes, active/passive.
Could we mirror a database from the production cluster to the DR cluster, for the sole purpose of allowing us to fail over to the mirror database, which would then be in its own DR cluster?
This sounds plausible in my head, but would this be a 4-node active/active/passive/passive cluster, or is it two discrete 2-node clusters?
Are there any gotchas that I should be aware of, particularly around licensing?
Any help would be greatly appreciated. I've tried searching the web for a definitive answe, but am struggling to find exactly what I need.
Thanks,
Innerise.
January 8, 2013 at 7:03 am
Innerise (1/8/2013)
Suppose we have a production SQL cluster, 2 nodes, active/passive.Also suppose we also have a DR cluster SQL cluster, 2 nodes, active/passive.
I can understand the provision of high availability on the Live site, i work with it myslef, but why on earth are you providing this on a DR system that will hopefully never get used???
I'm glad I'm not responsible for your IT budget spend :w00t:
Innerise (1/8/2013)
Could we mirror a database from the production cluster to the DR cluster, for the sole purpose of allowing us to fail over to the mirror database, which would then be in its own DR cluster?
You could, but again why the overkill on the DR site?
Innerise (1/8/2013)
but would this be a 4-node active/active/passive/passive cluster, or is it two discrete 2-node clusters?
Why, are the 4 nodes part of the same cluster? Are they even in the same domain?
Someone needs to take a step back, research and go back to the drawing board on this one. If you really want HA across the Live and DR site then a geo graphically dispersed (or multi site) cluster would be appropriate. Just be careful not to end up with a DR vote heavy cluster site as you'll likely have random failovers which won't make you popular.
-----------------------------------------------------------------------------------------------------------
"Ya can't make an omelette without breaking just a few eggs" 😉
January 8, 2013 at 7:19 am
The high level design has been scoped such that in the event that we move to DR, the (new) production environment will be in its own cluster ready to go. This is what has been agreed at the moment and there is budget for it. I guess it could be considered overkill, but is there such a thing when it comes to DR?
I should've mentioned that the two clusters are in graphically dispersed sites in the same domain. The mirror config will not use a witness server, so that we have full control over the failover to DR.
We're going down this route so that should we lose site A, we're not immediately exposed with a single production database at site B without any HA provisioning of its own.
Viewing 3 posts - 1 through 2 (of 2 total)
You must be logged in to reply to this topic. Login to reply