For the last couple of days I have been looking at the options that Windows Server 2008 Enterprise offers in terms of Clustering. I have never really looked at clustering before (apart from playing around with Hyper-V and clustering last week).
As we do not use any live Virtual systems here in the office I thought that testing and playing around with that would not be good business development so I looked at the option for File Share clustering and also printer clustering.
Up until now Clustering had been rocket science, you needed the same hardware and some would even say you needed sequential serial numbers on your kit, same BIOS revisions and even then it would still take a lot to get the cluster up and running. For my testing I had a Dell PowerEdge SC440 and a Dell Optiplex as the clustered servers and a Dell Precision Workstation running Windows Storage Server 2003 r2 (WSS 2003 r2).
Currently we have the following setup:
We have a single file server providing shares for all our data that is stored on the SAN. The problem we face is if the server fails then we will find ourselves With data in the excess of 1TB that we cannot access.
The server we have has full redundancy in it, redundant disks for the OS, redundant power supply redundant fans etc, but if we have an OS crash then the data that is connected to the server will be unavailable.
This whole thing raises concerns as the data that is on this box is critical to the functioning of the business and without it we would not be able to work. So I thought that I would look to see how I can setup a Failover Cluster in an iSCSI environment to test what would happen if a server did crash..So my testing setup is:
As this is a test environment the iSCSI is running on the same network card that gives the servers access to the LAN. Ideally you would have a dedicated network card for your iSCSI setup so that you can reduce the network traffic on your actual LAN.
For the test setup I created a 2 VHD's on the WSS 2003r2 box 1 x 100GB and 1 5GB then added them both into the same target group and then added the two initiators (server1 and server2) to the target group.
I then connected to the disks from Server2 using the iSCSI initiator. Once connected I formatted the disks and assigned the drive letters. Once that was done I ran the iSCSI Initiator on the server3 and connected to the disks.
I then followed this guide on: Configuring a Two-Node File Server Failover Cluster
Once done I could then access my failover Cluster from any PC or server on the LAN.
I then created a share on the Cluster and copied some data to it. Opened up a Word Document in that was stored in the cluster and then started working on it. I then simulated a failure by pulling out the power cable (some of you may have actually experienced this when BT have been on site).
Once I had pulled the cable out of the server that was currently the live node I noticed that the file view I had had frozen, however the word doc was still open and I was still able to edit it. after a few seconds I hit the close button on the doc and it prompted me to save changes so I hit yes, and it saved.
So in light of this I thought if it works for iSCSI it should work for a SAN. Although I have not configured it on the SAN yet I would imagine that it would be in the same way I would just zone in 2 servers to a zone rather than just one. like you do with the iSCSI. so I would imagine it would look something like this:
The current Zone I have setup for the File Server on the SAN is:
Switch1:
FileServer HBA_Port1
SP_A1
SP_B1
Switch2:
FileServer HBA_Port2
SP_A2
SP_B2
However I guess for the Cluster Setup I would need the second server listed in the zone too.