SOIC watches I/O latency across the datastores shared in a VMware Cluster and enables each VM it's fair share of disk resources in times of contention when a certain latency value is exceeded (30ms by default). This ensures one disk intensive VM doesn't impact other VMs disk performance. You can also "tweak" and say some VM need priority access in times of contention as well (at the expense of the other vms on that datastore though).
· Storage I/O Control is supported on Fibre Channel-connected and iSCSI-connected storage. NFS datastores and Raw Device Mapping (RDM) are not supported.
But check out these links for more info:
http://www.yellow-bricks.com/2010/09/29/storage-io-fairness/
http://geeksilver.wordpress.com/2010/07/28/vmware-vsphere-4-1-storage-io-control-sioc-understanding/