ROFL..I'm sorry but with all the CEF IOS CatOS SUP1 MSM VLAN stuff it just
looked funny to me...I'm easily amused ;) I guess it's just the way every
other word came out an acronym...heh.
----- Original Message -----
From: "dre"
To:
Sent: Tuesday, June 26, 2001 2:49 PM
Subject: Re: Catalyst 6500 Confusion [7:9983]
> ""John Neiberger"" wrote in message
> [EMAIL PROTECTED]">news:[EMAIL PROTECTED]...
> > Okay, I'm thoroughly confused. I'm doing some research on this platform
> > because we'll be purchasing one later this year or early next year. The
> > documentation seems to be quite murky and the more I read, the less
> > clear the picture is.
>
> The entire Catalyst architecture is strange and piecemeal. I understand
> your
> confusion and frustration.
>
> > I see a few different issues that I need help with. The first is
> > software. Which one to use? CatOS? Supervisor IOS? If I can only do
> > CEF with IOS, then if we get CatOS are we totally hosed and limited to
> > software forwarding of interVLAN traffic?
>
> The original SUP1 MSM architecture of the Layer 3 Catalyst 6500 is
> different than the SUP1A MSFC architecture is different than the SUP2
> MSFC2 architecture. You can use the MSFC2 on the SUP1A, but you
> don't get the advantage of CEF in Layer 3 switching like you would with
> the SUP2 MSFC2 combination. SUP1A has to use MLS. SUP2 has
> to use CEF.
>
> In the SUP2 MSFC2 architecture, CEF is downloaded from the MSFC2
> to the PFC2 (another card on the SUP2), so CEF now forwards in
> hardware. This happens whether you are running CatOS (Hybrid) or
> Sup IOS (Native). If you are using DFC-enabled linecards, the CEF
> table is also downloaded to each DFC (there can be up to two per
> linecard, depending on which ones you are using). It is actually even
> more complex than this, but I don't want to make your head spin.
>
> The only real difference between CatOS and Sup IOS is support and
> features.
>
> CatOS + MSFC IOS is currently the software for the SUP2 MSFC2
> architecture. This is changing in the next three months. Everything
> currently only works properly for SUP2 MSFC2 under CatOS.
>
> But three months from now, Cisco is changing directions with the product
> and moving completely to Sup IOS. All future work will be done for Sup
> IOS, but it is not available fully yet. The newest features are being
> developed for Sup IOS like EoMPLS (which Cisco demonstrated on the
> Catalyst 6500 / 7600 OSR at SuperComm). I believe that the Catalyst
> 6500 / 7600 OSR is the only product currently supporting EoMPLS (or
> any of the AToM technologies), even though it doesn't yet support
> MPLS-TE or MPLS-VPN. More details on EoMPLS are available at
> http://www.cisco.com/warp/public/cc/pd/rt/7600osr/prodlit/emp76_tc.htm
>
> > I see that the MSFC2 can do 650 Mbps+ of software-based forwarding. Is
> > this only referring to first-time routing of packets when no MLS entries
> > are present in the cache? If there are cached entries are they
> > hardware-forwarded or are we still limited by the performance of the
> > MSFC?
>
> MSFC2 only does forwarding when there are no MLS entries in the
> cache, yes, but this only would occur on the SUP1A architecture. With
> SUP2, the MSFC2 would only do forwarding when there are no CEF
> adjacencies or when packets are not-cef-switched (first packet generally).
>
> The benefit of having the Catalyst 6500 would be to use MLS with SUP1A
> whenever possible and CEF with SUP2 (or to the DFC-enabled linecards)
> whenever possible. This maximizes performance, because it's all done in
> hardware forwarding.
>
> > We have a lot of interVLAN traffic and my worry is that as traffic
> > volumes increase that our performance will suffer due to restraints
> > present in the MSFC.
>
> As long as everything stays in MLS (SUP1A) you will be ok. There is
> a hashing algorithm problem that causes a maximum of 128k destination-only
> based flows, but many people have found that realisitically this number is
> much lower due to a hash collision that will result in packets forced up
to
> the MSFC (1 or 2, doesn't matter). More details can be found at:
> http://www.cisco.com/warp/public/473/35.html (at the bottom of the
> page). Setting destination-only based flows (the default) and by tuning
> with MLS fast aging, you can possibly acheive up to 128k flows in MLS.
> The MSFC will handle a lot of packets, still, however the performance is
> not optimal compared to MLS.
>
> Even better would be to use SUP2 MSFC2 with DFC-enabled linecards.
> They do not suffer from the problems of MLS and things like OSPF ECMP
> work a lot better using CEF. The performance of CEF is much better.
>
> Also, SUP2 MSFC2 with Sup IOS seems to be the product's overall
> direction.
>
> > Without IOS and CEF, once the MLS cache is populated, aren't any
> > further packets hardware switched? If that's the case, I don't see much
> > need to get the Sup IOS. I seem to remember someone complaining about
> > the Sup IOS but I don't remember the details. Hmm...I suppose I ought
> > to check the archives, huh? :-)
> >
> > Anyway, any help anyone could offer would be greatly appreciated!
>
> CatOS is what most people are comfortable with. It also support the
> most features currently, but probably won't support any new features.
> I don't know of any currently installed Sup IOS installations, and I have
> not used the code myself, however, there is no arguing that in three
> months and down the road, Sup IOS will evolve as the preferred
> software for the Catalyst 6500 / 7600 OSR architecture.
>
> -dre
Message Posted at:
http://www.groupstudy.com/form/read.php?f=7&i=10000&t=9983
--------------------------------------------------
FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html
Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
