January 26, 2009 at 12:54 pm
I was reading a white paper entitled: Microsoft High Availability Overview. It mentions that SQL Server Always On Technologies which include Database Mirroring and Failover Clustering can be used in combination. I’m not too sure as to how that happens. If I had a 2 node Windows 2008 cluster with an active/passive configuration running SQL Server 2005 and a SAN back end, I understand that in the event of a hardware failure, the failover clustering takes over and control is given to the passive machine. I’m somewhat confused as to how database mirroring figures in here. In database mirroring, two live instances of SQL server is a requirement in each node. If the SAN fails then what good is the mirror then? This is assuming that there is no SAN redundancy implemented.
I always thought that the two were exclusive. I’m tasked with investigating this. I have virtually no clustering experience as my former employer used DB2 and ORACLE for their tier 1 database applications. Now, I am in a shop that exclusively runs on SQL server.
Any guidance from the experienced is appreciated.
Thanks in advance.
January 26, 2009 at 4:25 pm
You are correct that if you didn't have SAN redundancy and you were to lose the SAN mirroring would not help. So, my belief would be that the document was trying to say that you could use clustering to handle the server hardware failure. Mirroring would allow for a more robust Disaster Recovery solution in that you could be mirroring to another site where the server would be using different storage. I haven't read the document though. If you post the section you are referring to we could read this and probably get a better idea of what you are referring to.
Thanks.
David
@SQLTentmaker“He is no fool who gives what he cannot keep to gain that which he cannot lose” - Jim Elliot
January 26, 2009 at 4:47 pm
Something else to keep in mind. If you are using a cluster to handle a server failure and a mirror to a separate server (or cluster) using a separate storage system that you may find that you will fail over to the mirror during a server failure. The failover to the mirror will occur faster than the failover to the clustered server. This, of course, assues that you are using automatic failover for the mirror.
This is based on much of the reading I have done as we look at the various DR possibilities here where I work.
January 26, 2009 at 6:25 pm
From experience with clusters and DR the one thing that easily forgotten in this is Backups. Because sometimes the thought process of the cluster/db's are mirrored or what ever, backups can easily be forgotten in the BIG picture. I currently have about 6 active/passive clusters and not only are they mirrored they too are backed up. Just make sure you dont depend on clusters for the wrong reasons..they may be ok to keep you up and going but dont forget the backups. But not sure if you are using the mirrors for the DR to the cluster while you restore from the backups. As you see this easily becomes somewhat tricky just make sure you keep it all straight in your mind. 🙂 Hope this gives a lil insight.
-Darryl
DHeath
January 27, 2009 at 6:31 am
Dave, Lynn and Darryl,
Thank you all for the insight. It is appreciated.
Dave,
Here is link to render the document in HTML:
Regards.
Viewing 5 posts - 1 through 4 (of 4 total)
You must be logged in to reply to this topic. Login to reply