Some medianet questions:
(1) Is support planned for Sup720 (C)? If so, when is it due to be shipping? It would seem most of the features require code porting more than explicit hardware support, and many of our Sup720 / 3C customers aren't yet ready to got Sup2T or Nexus (I expect 6509E w/ Sup720 C in closets to persist for quite a while, for example).
(2) I see one partial limitation to the interesting announcements is they require recent hardware and current code. Many of our customers are still buried in work, and generally behind in terms of installing security upgrades or new code. And may take 3+ years to get to current hardware. Ok, so medianet will work with the equipment that supports it, so network core and ISR G2 gets you a pretty good starting point -- most sites have gone to ISR G2 or ASR and it's their closets that lag equipment-wise.I guess there's not much you/Cisco can do about that. Any ideas? Obviously medianet is an incentive to upgrade to newer hardware, but given the cost, it is unlikely to cause that to happen very much sooner.
(3) Lack of Nexus support seems only a concern for signaling to and streaming video sources in the datacenter. Except what about conference sessions with an MCU in the datacenter? Is a Nexus solution / code for medianet coming (I think that was mentioned as a "yes" but with no indication of how soon.)
(4) NBAR on switches and scaling always worried me since the PISA chip (which I viewed as an order of magnitude too slow at the time, it needed 20 Gbps for 2 x 10 Gbps uplinks, which is what I've been designing for big closet switches for quite a while). So I in effect removed NBAR from my mental list of "useful QoS tools" at that time. (I very much want uniform configuration across devices, supporting this port or this device type does something different is unsustainable.) It sounds like medianet uses NetFlow to capture data stats, which means near-wire-rate recognition of packets. So does medianet provide higher speed QoS classification now, or should I still be thinking "match application" = "match protocol" = NBAR = *SLOW* speeds, WAN edge routers only? Since you support it on Sup2T, I'd hope it can do wire rate -- does it, and if so can you give a vague indication of what makes it so much faster than historical NBAR? Can you confirm / deny my NetFlow theory, and while you're at it, provide info about any medianet-related caveats (as in "if you do THIS, it'll bog things down").
(5) Any other medianet scaling / performance issues or factors to be aware of? It doesn't do any of us any good to pitch it to a customer as the greatest thing since canned beer, and then have it fall flat on its face due to hitting some corner case or walking into a performance issue.
(6) Licensing or other costs? The presentation talked about the IOS feature sets required, is that the only costing issue? Well, of course if you want to use Prime, there's purchase cost for that, per-device, licensing, etc. Is there anything else to factor in when explaining costs of deploying medianet to the customer?
Thanks!
With Eric Yu
