No problem, I'll definetly let you know how I make out. One thing I know for sure, is I will definetly be implementing RDMs with the ESX iSCSI initiator. From what articles and whitepapers I've read, RDMs have some advantages over the MS iSCSI. There is a ton of stuff if you google "rdm vs ms iscsi". Here's just one blog I read http://blog.core-it.com.au/?p=419, but it lists some good pros and cons of each. The biggest thing for me is the compatiblility with SRM. We will be looking to implement SRM over the next year or so, so for scalability I'm going RDM. Plus the performance benefits seem to favor RDM. Which makes sense. When you think about it, the MS iSCSI initiatior is inside the guest VM. Which means your going through the vNICs and all the virtual overhead. RDMs, at least how I understand them, are a bit different. It is using the physical NICs on the hosts to connect a "virtual disk" RDM to the guest VM, allowing the VM, what I'll call, 'almost direct physical access' to the LUN on the SAN.
In my opinion using the MS iSCSI is simpler to implement. You install it, point it to a target, and your pretty much done, aside from any multipathing you choose to do. Using the ESX iSCSI I dont think is more difficult, but I think it is a bit more involved. There is a little more to setup. And using an RDM with a VM that lives in an NFS datastore can be done. It is even vMotion capable, as long as your RDM mapping file is on block based storage like FC or iSCSI.
So, all that being said. So far, I have enabled the iSCSI initiators on my ESXi hosts and pointed them at my SANs storage IP addresses. Created an iGroup on the netapp side that includes the initiator names from my ESXi hosts. I created a volume called "rdmmap" and a LUN inside it called "rdm". I added this LUN as shared storage to my ESXi hosts. Then I created two new volumes called "db" and "logs". I used Snapdrive to appropriate the LUNS in the proper volume with the iGroup I created earlier. These LUNs are now seen as local drives on my Exchange server. Now when I have some downtime this weekend and next I'll migrate the EDBs and all to the new drives on the SAN and configure SME.
As far as backups go, I created a new NFS volume just for the C:\ drive of my Exchange server. 99% of my VMs are in NFSPROD, which are snapped via SMVI with vmware consistent snapshots. Exchange is in a seperate NFS volume called EXCHNFS.
Even though I am not using the MS iSCSI initiator, I am having some issues with getting a VMware consistent snapshot of Exchange. The snapshot pauses at 95%. I've let it sit for at least 5 minutes and nothing happens. I have to power off the VM for the snapshot to complete. I don't know for sure if it should be able to complete or if there is some know issue in snapshotting a VM with RDMs or with SnapDrive, but since Exchange is in EXCHNFS and I created a seperate SMVI backup job to just do a NetApp snapshot (no VMware Snapshot), like your doing on all your VMs, Marc. This is my Exchange "hot" backup. Then I plan on scheduling vCenter tasks to shut down Exchange at a certain time, having a "cold" SMVI backup job snapshot it, then a vCenter task to power it back up. Since I can afford the 30 minute downtime, this seems like the best and safest option for me. And also, in addition to SMVI, SME will be snapping the databases and logs (I'm leaning towards every 6 hours).
As far as the "quescing" check boxes, I'm not entrely sure, but I'll see if I can find anything. Let me know if you have any other questions about the ESX iSCSI or anything else. And thanks again for your help. I'll be sure to post a final update when I get my DBs moved over and everything is completed.
