I'd like to know why the Fibre Channel community has not standardized a multiple-link scheme yet. I can take two $10 NICs and use software to aggregate them together. Switch-based trunking has been around for years and even the IEEE has a standard for it (802.3ad). I think the storage industry is way behind the curve here.
Is there a simple answer as to why it hasn't happened yet? When Brocade pushed FSPF forward, I thought the floodgates were opening, but I've been sadly disappointed.
It's difficult for me to answer on behalf of a technical body (SNIA) or an industry to which I have limited exposure. However, what I can tell you is that this is a harder technical problem than it appears. Fibre Channel low-level protocols assume a lot more about the reliability of the link than TCP/IP. So there's no provision for out-of-order delivery (which is not normally an issue with storage I/O). Also, if the reason for trunking is to increase single-stream bandwidth, host-based striping on independent Fibre Channel paths accomplishes that.
Since there is limited value, and the problem can be solved another way, there probably isn't a lot of impetus being placed on solving the problem. That may change over time as other hard SAN problems are solved. But I suspect that this one is a little further down the priority list.
Hope this helps.
Editor's note: Do you agree with this expert's response? If you have more to share, post it in one of our .bphAaR2qhqA^0@/searchstorage>discussion forums.
This was first published in April 2002