This may be due to the fact that they are in VMs.
If the tool is measuring actual I/O on the card, some of this will be reduced due to the fact that you are now taking advantage of a logical network built into your host to talk to other VMs on the same host.
Moving from local disk on a physical server to a VM makes your disk I/O occur on the SAN, between .vmdk files. This reduces the disk I/O partially because of the improved capabilities of the disks in your SAN, and partially because you're now transferring (potentially) from file to file rather than from disk to disk.
Typically, most of the network IO would be between VMs and systems outside of the virtual infrastructure.
Lowering network traffic by almost 50% to account for inter-VM traffic doesn't make much sense.
Only traffic between VMs on the same host and vswitch would not generate any external I/O.
Same for disk IO: most of the data written to or read from a VM will originate from or get sent to an external source, not the VM itself.
File copies between disks in a single VM (or on a physical server actually) would not occur very frequently.