Even with a single chip architecture the overall scale performance is WAY better than Sup720. Hell, even RSP720 was a huge improvement in scale
I know the question was specifically about CPU but Sup2T is also a different forwarding ASIC allowing it to do natively things Sup720 couldn't, like VPLS and EVC I would agree that Sup2t wouldn't be my first choice in ISP Edge. From Cisco, ASR9k or ASR1k depending on bandwidth needs. -Pete disclaimer: I work for Cisco. On Mon, Dec 9, 2013 at 10:34 PM, Jimmy Hess <mysi...@gmail.com> wrote: > On Fri, Nov 29, 2013 at 12:02 AM, Mikael Abrahamsson <swm...@swm.pp.se > >wrote: > > > [snip] > > > +1 for MX or ASR 9000. > > > > Cisco ASR 9000, Juniper MX, Huawei NE40E, Alcatel-Lucent 7750, those > kinds > > of routers are the ones I hear people using. Some go for the new Sup2T > for > > the 6500, but I don't know how much more CPU it has compared to your > > SUP/RSP720, perhaps someone else knows? > > > Cat6500 Sup720 was a platform that used two separate processors; 1 Switch > Processor CPU at 600mhz managing Layer 2 services, and 1 Route processor > CPU at 600MHz on the MSFC to run the Layer 3 services. these were MIPS > CPUs --- sr71000. > > Cat650 Sup2T is shown as a single Dual core, 1.5GHz per Core cpu. There > is one processor stack on the 2T, instead of two separate CPUs; since > route processor and switch processor are now combined into one shared > processing unit under the new "merged" architecture that runs only one IOS > image, that controls both RP and SP features ---- Layer 2, Layer 3, and > management services do not run on separate processors, with their own > separate hw anymore. > > > So the CPU is beefier --- but it is also now shared by multiple functions > that previously had separate, isolated processing from one another. > > I believe the Sup2T are using a E500 PowerPC chip. > In any event, neither old nor new are based on x86 architecture --- keep > in mind, that comparison of MHz or GHz CPU frequency rates is only > meaningful within the same CPU architecture. > > There are not significant increases in FIB TCAM, or other important memory > capacities from RSP720, that you would expect to need for scalability to > larger tables. > > > Even with 2T I would still describe the 65xx as largely a great switching > platform, for 10/100/1000 aggregation -- due to limited chassis > bandwidth: its days would seem to be numbered once desktops are sporting 10 > gigabit links: definitely not (IMO) the best hardware router platform > for carrying large routing tables at the ISP edge, anyways. > > > > > Mikael Abrahamsson email: swm...@swm.pp.se > > > -- > -JH >
