I would have classified the ISR 445x-X range of routers as hardware-based routers, based on the different hardwares (both memory and "CPU"/forwarding-engine) for control plane and data-plane. On the other hand, I understood that the ISR 443x range are software-based routers (same CPU/memory for control and data plane).
On Thu, Jan 15, 2015 at 2:20 PM, Phil Mayers <[email protected]> wrote: > On 15/01/15 12:26, [email protected] wrote: > >> Interesting your distinction because Cisco says that the 4451: >>>> "The product???s innovative hardware design splits the control and data >>>> planes between two multi-core CPUs??? >>>> >>> >>> Indeed, that's where my definition would not trivially apply either way >>> :-) >>> >>> (And given how fast multipurpose CPUs have become, we see this in other >>> areas as well - like "SSL offloading PCI boards" that you could add to >>> a web server for decent HTTPS performance... most of that has just plain >>> disappeared with modern CPUs...) >>> >> >> For me the relevant question is often: Can the box handle line rate, >> or close to it, with minimum sized packets (and any access lists, >> services, QoS etc that you need)? If it can, great - otherwise you're >> opening yourself to a DoS attack at some point. >> > > I tend to agree. I like predictable performance that degrades sensibly, > not suddenly crossing a threshold and failing to work at all. Whether it's > FPGA, TCAM or CPU and fast RAM is in theory irrelevant. > > In practice, vendors tend to be rosy about their performance, and the > underlying architecture can indicate the likely performance window ;o) > > _______________________________________________ > cisco-nsp mailing list [email protected] > https://puck.nether.net/mailman/listinfo/cisco-nsp > archive at http://puck.nether.net/pipermail/cisco-nsp/ > _______________________________________________ cisco-nsp mailing list [email protected] https://puck.nether.net/mailman/listinfo/cisco-nsp archive at http://puck.nether.net/pipermail/cisco-nsp/
