A. CSVs allow all nodes in a cluster to access storage simultaneously, even though one node still "owns" the disk. The node that owns the disk is called the coordinator node and has the disk mounted locally. All nodes in the cluster have to forward any metadata type changes to the coordinator node for actions such as file size changes or properties, but the other nodes can directly perform I/O level actions on the disk. Direct I/O traffic should be sent directly over the iSCSI network.
In some circumstances, this direct I/O isn't possible. For example, file copy operations that use Server Message Block (SMB) consist of numerous types of I/O that aren't compatible with direct I/O. SMB copy authorizations use redirected I/O, so all of the I/O is sent to the coordinator node. Data sent to the coordinator node uses the NetFT adapter, which is probably not bound to your iSCSI network, so you'll see the traffic going over a different adapter. This is why it's important when using CSV to have a high speed network that can be used for the cluster communications.
Note that redirected I/O is also used for other actions such as backing up, defragmenting, or running Chkdsk on a volume.
For more on networks and clustering, see the following video:Related Reading:
- How do I create a failover cluster in Windows Server 2008?
- Is a Windows Server 2003 and Windows Server 2008 mixed-cluster possible?
- How can I add disk resources to a Windows Server 2008 cluster from the command line?
- How do cluster shared volumes work in Windows Server 2008 R2?
- How to Manage Failover Clusters in Windows Server 2008
- Configure Windows Server 2008 Failover Clustering from the Command Line
Check out hundreds more useful Q&As like this in John Savill's FAQ for Windows. Also, watch instructional videos made by John at ITTV.net.