01-04-2013 06:09 AM
We have an IBM blade linked to an ethernet fabric formed by two VDX6730-76 (NOS 3.0.1).
We would like to setup FCoE over VLAG towards each blades but documentation and google fu seems to say that it's not supported.
Can someone confirm ? If not, can someone share configuration examples ?
01-04-2013 06:22 AM
--->>>Can someone confirm ? see attachment
vLAG and FCoE Traffic
The vLAG feature does not support FCoE traffic, due the requirement of in order delivery of frames for FCoE, which is
not provided with vLAG.
01-04-2013 06:27 AM
That's what we read and what we were told by our SE.
Still, if I refer to the "Brocade VCS Fabric Technical Architecture", one can read at the beginning of VLAG chapter :
"Virtual LAG (vLAG), an innovation that is included in Brocade VCS Fabric technology, extends the concept of LAG to include edge ports on multiple VCS switches. Edge ports in a vLAG support both classic Ethernet and DCB extensions. Therefore, any edge port can forward IP and FCoE traffic over a vLAG as shown in Figure 11."
Am I understanding this incorrectly ?
01-04-2013 06:32 AM
--->>> Am I understanding this incorrectly ?
--->>>Therefore, any edge port can forward IP and FCoE traffic over a vLAG...
with other word, that mean edge port can "forward" in not the same can handle FCoE Traffic over vLAG.
01-06-2013 03:42 PM
i was looking for other DOC, and found this interesting PDF.
have a look expecially on page 3, this explain how FCoE can be deploy over vLAG.
is a bit courious, other DOC say ...vLAG feature does not support FCoE traffic,
let me know if this is a solution you was looking for.
01-09-2013 05:17 AM
That is the DOC (or at least an old version of it) I was refering to.
One of my storage colleague was able to config FCoE over VLAG on our platform.
I am curious to know Brocade's official position on this subject as we have heard everything and it's contrary.
Is FCoE over VLAG supported or not ?
01-12-2015 08:13 AM
FCoE over vLAG is not supported in Brocade VCS fabrics. Fibre Channel has radically different HA requirements than IP. When we do LAG in the IP world we are essentially creating a single virtual IP interface with a single IP address. Fibre Channel requires two independent initiator devices, each with its own FCID (24-bit Fibre Channel address) and WWN, which are typically connected to two isolated (air-gapped) fabrics. You map the same LUN from the storage device through its redundant controllers to the two initiators and then use a multi-pathing driver (MPIO) to present a single virtual LUN (that is backed by the two redundant LUNs which are the same) to the OS so that it doesn't see the same device twice. Thus, each initiator (CNA port) needs to obtain its own FCID, and for that reason each one has to be independent and they cannot be LAGed together.
You can combine both features though. If you have a server with a dual-port CNA you can connect each port to a different VDX switch in a VCS fabric, form a vLAG for IP traffic and use each independent link for FCoE traffic (by enabling FCoE at the individual vLAG member level and not at the port-channel level).
FCoE over LAG (not vLAG) was introduced somewhere in NOS 3.0.x to enable the connection of FIP Snooping Bridges (FSBs) to a Brocade VCS fabric and to run FCoE between the FSB and a VDX switch with redundancy of uplinks between the FSB and the VDX. But this allows to connect an FSB to a single VDX switch and have a "regular" LAG between them.
The attached presentation my shed some light.