12-08-2010 04:45 AM - edited 03-01-2019 09:45 AM
Hi:
The UCS ecosystem leverages a port aggregation solution to chassis I/O, namely, the FEX modules.
The FEX modules are not fully featured switches. Nor do they possess any forwarding policy intelligence at all. Instead, the FEX modules deploy a "pinning" approach in which downlinks (those that face the blade server's NIC's, LOMs, mezzanine cards) are mapped to an uplink port (those that face a 6100 Fabric Interconnect) to form what can be described as an aggregator group.
The result is a simplified approach to blade I/O in which the traffic patterns are predictable and failover is deterministic. Moreover, there is no need to configure STP because the ports are uplinked in a manner as to preclude any possibility of a bridging loop.
This having been said, is there some merit to the argument that this port aggregation design places a hole in the middle of a QoS deployment since the scheduling of packets on the uplink ports facing the 6100 Fabric Interconnect is not performed in a manner that recognizes priority?
To elaborate a bit more, one can have a VMware deployment and leverage NetIOC or perhaps configure QoS on a 1000v switch (whose uplink ports are mapped to a port on the Palo VIC) and configure QoS on the VIC, and then on the 6100 Fabric Interconnect. But, since the FEX is not scheduling traffic to the 6100 Fabric Interconnect according to any priority, the QoS deployment has a hole in the middle, so to speak.
Thoughts?
Solved! Go to Solution.
12-08-2010 09:01 AM
Hi,
I have posted a response to your question here:
http://bradhedlund.com/2010/12/08/cisco-ucs-fabric-extender-fex-qos/
Hope this helps.
Cheers,
Brad
12-08-2010 06:03 AM
Have you reviewed Brad's article on UCS QoS? http://bradhedlund.com/2010/09/15/vmware-10ge-qos-designs-cisco-ucs-nexus/ It may answer some questions. Based on his article, it looks like you can do some bandwidth manipulation at the blade level.
I see why you think there may be a "hole" at the FEX level. It's a good question. I wonder what would happen if you had two blades on the uplink, both with the highest class of service. A 10GbE port is still a 10GbE port. Do they drop down to 5Gb each or does some other mechanism come into play?
Adam
12-08-2010 07:44 AM
So I posted your question to Brad and here is his response:
"In short, the FEX pushes congestion back to the adapter (PFC), where the adapter decides traffic scheduling"
Adam
12-08-2010 08:28 AM
Thanks, Adam.
Yes, I am aware of Brad's blog. Read it all the time. And I read the QoS section.
I cant say I understand his answer to this, though...wish he would elaborate. I also posted the question on his blog a few nights ago...not sure if hes answered me yet. Will check.
http://bradhedlund.com/2010/09/15/vmware-10ge-qos-designs-cisco-ucs-nexus/
12-08-2010 09:01 AM
Hi,
I have posted a response to your question here:
http://bradhedlund.com/2010/12/08/cisco-ucs-fabric-extender-fex-qos/
Hope this helps.
Cheers,
Brad
12-08-2010 10:23 AM
Brad, good stuff! I read your blog all the time. I especially like the 10-part Best Practices video series. You pay excruciating attention to detail, which is what any engineer should do. I hope Cisco takes care of you, if you know what I mean. :-)
Anyway, I read you answer quickly. I'll try to digest it as I meet some clients. I may get back to you on it.
If not, let me say thnk you now for your time and answer. Much appreciated.
And by the way, I only ask these types of questions so that I can fully understand the solution, not to say "gotcha." Its necessary to play devil's advocate sometimes - you know your clients will!
Discover and save your favorite ideas. Come back to expert answers, step-by-step guides, recent topics, and more.
New here? Get started with these tips. How to use Community New member guide