Recently I've replaced a 7200 NPE-G1 router with so called better processor engine, NPE-G2. The router was handling something like 200-300mbps traffic on each 3 onboard interfaces, doing some inter VLAN routing, NAT, route-map and a small single area OSPF. The CPU usage at peak time was %70. We decided to replace it with a better NPE (NPE-G2) but surprisingly we didn't see any better performance from the new NPE-G2! The NPE-G1 didn't use even the MPF (Multi Processor Forwarding). As i checked the data sheets of both NPEs, it's clear that everything has been doubled from memory, PPS and even the CPU speed (700MHz to 1.67GHz!!!). I've seen some postings that even the onboard Gigabit ports cannot forward 400mbps as before (NPE-G1). Can any one explain what is going on with new CISCO products???
If your CPU utilization was 70% before the upgrade, then you weren't reaching the limits of your NPE-G1. Since this is the case, the upgrade to the NPE-G2 wont help except to permit future traffic growth.
Unless you're seeing 'ignored' errors on your interfaces, packets aren't being dropped at the 7200, and it is forwarding everything that comes in.
Is the CPU utilization processing the same amount of traffic lower now on the NPE-G2? If so, you know it is doing its job.
The NPE-G1 was fine and we do not receive any error on the interface (ignored...) but we were a little worry for times that we are under denial of service attack (e.g. High PPS). The problem is that the CPU utilization of both NPE-G1 and NPE-G2 are the same in situation that configuration and traffic amount are the same!
in practice the CPU utilization, is not a good parameter to tell how much room is left for processing.
It also need to be read correctly, eg. how much time spent at interrupt time, how much at process time (the first two values, separated by /).
I've seen router running perfectly fine with 90% CPU utilization, others dropping like crazy with just 30%. All depends by the architecture, traffic pattern, and too many other things as well.
One thing I can tell you in total honesty, if Cisco says that NPE-G2 does 2 million PPS Vs 1 MPPS for NPE-G1, we can argue that the measurements are taken in ideal lab conditions, argue about what packet size, etc. but the ethic of the company is never to deceive customers, be reassured!
I agree with all you said, but what i expect after changing the NPE (from G1 to G2) was to see a little change in CPU usage of the box when handling the same amount of traffic but nothing happens! That's the strange part.
We are pleased to announce availability of Beta software for 16.6.3. 16.6.3 will be the second rebuild on the 16.6 release train targeted towards Catalyst 9500/9400/9300/3850/3650 switching platforms. We are looking for early feedback from custome...