I think it’s overly simplistic to redact the problem to bandwidth. Extra bandwidth helps, but the underlying problem from my point of view, has been the lack of team integration between server / storage / networking at a functional level as well as a technological level. Let me generalise a bit.
The Network team tended to the view that the visibility of the “server” was the connection of each adapter – therefore a separate port for each function gave visibility and control of the network edge. Of course, this isn’t particularly true but access to vCenter was a problem and didn’t have good networking services anyway.
The Server team tends to the view that the network performance is unknown and thus having lots of ports provides enough performance and some certainty that performance wouldn’t suffer. This, also, isn’t particularly true and lack of understanding of network technologies and integration meant it was common.
The Storage team likes being isolated, and continued with FibreChannel because it suited their operations. FibreChannel is massively over engineered for most applications and uses, and wasted a lot of money. Storage networking uses relatively little bandwidth but does have specific performance needs.
What I think has changed is that the industry drive to efficiency brought cross-discipline exchange of ideas that means new methods can come to the surface. Storage can now converge onto Ethernet connection and shared port usage is acceptable because visibility, performance, QoS and serviceability issues are addressed (or are being addressed).
It’s more complicated than this, of course, but hopefully the point I’m making comes through.
The nature of 802.1Qbh/Qbg is that another form of software based networking is possible — the server and switch can exchange configuration data and status via standard data interfaces. This leads to visibility of all layers for server/network/storage needs and solves the challenge around isolated data sets. This data exchange would be approximately similar to VAAI/VASA is concept and imagine VMware/Xen virtual switch software could query the LAN Switch for status , configuration and capability data.
I suggest that reading about Cisco’s VM-FEX technology, which is a pre-standard implementation of the 802.1Qbg technology, might be useful if you would like some insight into the possible future.
Of course, OpenFlow is another answer to the same challenge but it’s not really developed enough to comment further at this time.