Take the 2-minute tour ×
Server Fault is a question and answer site for system and network administrators. It's 100% free, no registration required.

I'm aware that you can't mount a single LUN to two read/write NTFS iscsi volumes without massive issues unless you are using clustered servers. However, I'm curious about mounting 1 iscsi lun read/write on one server (say, a database server) and the same iscsi lun as read-only on another server (say, backup server) to facilitate easier backups as a interim solution.

Now, this is only for the database .bak files from SQL Server 2008 - the reason being I'm having other network issues that can't get resolved yet (network team is different side of the company, we don't have access to fix our issue) causing backups to take 20-25hrs. However, since they both have connections to the same iscsi SAN, I figured this would make things better.

I haven't tested this on my production system yet but haven't seemed to have problems running a similar test on my LAB - however with file-sizes much smaller (2gb compared to 800gb), I wanted to see if anyone had experiences in doing this and if it'll work or not.

Edit: Just for reference, both servers are Server 2008 R2 (SQL server is Ent, Backup is Std) connected to an HP P4500 (Lefthand) iscsi SAN.

share|improve this question
    
so just copying off .bak files takes 20 hours? Yet the server has no production performance issues? –  Bret Fisher Mar 5 '12 at 22:17

2 Answers 2

I don't see why that wouldn't work. What I'd recommend doing instead would be to take a snapshot of the production LUN, then presenting that snapshot to the backup server. That way if the production LUN changes the backup server will still see the consistent view of the data. Then when the backup is done delete the snapshot and the next day start the process over.

share|improve this answer
    
yea the hard part is automating that so that you have the latest SAN snapshot mounted. You also need to ensure the SAN is telling SQL to properly snapshot the DB's so you get a complete copy rather then just a raw copy of open DB/log files. –  Bret Fisher Mar 5 '12 at 22:36
    
That all depends on how much scripting the array supports. Sometimes it's pretty easy some times it's a lot harder and you have to be a lot more creative. –  mrdenny Mar 6 '12 at 20:09

I have also faced the same issue. SAN team published single lun on 2 servers, after we faced the drive corruption issue & even the fiber connection was fluctuating. After that we unmounted the drive on one server & unpublised that lun from that servers. Then it's started to work on first server after running checkdisk command. Eariler chkdsk command was not getting completed when the same lun was publised on both servers.

share|improve this answer

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.