Depends on what you're doing feature wise and whether you can withstand some (minor) packet loss.
We have 7200s doing PPPoE termination that get close to 500 out/250 in (about 55k/45k packets) on one interface (PPPoE) while at the same time pushing the same traffic 250 out/500 in (uplink). At that point the CPU is at roughly 80-90%. At that point you have input (overrun) errors however and some minor packet loss, you don't notice it so much with Internet traffic, but if you're doing Multicast/VoIP traffic or such you will notice it... For sensitive (VoIP/Video) with at least some basic QoS traffic a 200/200Mbit mix is more realistic. If you're doing NAT, access-lists, uRPF, MPLS that number will be lower and depending on what mix you have of those even substantially lower. I had a funny issue after upgrading to SRE the other day which turned out to be a simple scheduler misconfiguration..., a fellow engineer had accidentally reversed the allocation values on some of the boxes, alloting a lot more time for processor tasks than actual packet processing (in fact it was set as 800/3200). This turned out to be a much bigger problem under SRE than it was under SB... I've found 20000/1000 scheduler allocate values can work for a box that is spending 95% of its time pushing packets and very little time doing protocol (OSPF/BGP/LDP) work and that you don't have to work on a lot (if the router is busy with those values the telnet/console session is quite sluggish). 3000/1000 is a more realistic value for a box that is doing a mix/match of protocol and packet pushing work and makes the box very workable, even under heavy packet load. If the packet processing value is too low it will cause (as I saw the other day) input ignored errors/no resource errors on both the NPE-G1 and NPE-G2 even under very low load. Kind regards, Sibbi >From: [email protected] >[mailto:[email protected]] On Behalf Of Scott Granados >Sent: Wednesday, June 22, 2011 1:55 PM >To: [email protected] Providers >Subject: [c-nsp] Average throughput of a Cisco 7200 VXR G1 gig interface? > >Hi, I have a basic question about the Cisco 7200 VXR G1 gigabit >interfaces? > >What do people find is the average throughput of the gig interfaces? >Assuming normal packet sizes and standard traffic how much milage do you >really get out of these? I'm having an issue pushing past 110 megabits in >one direction and 15 megabits in the other over one of these gig >interfaces >and wondered what I could realistically expect under real world >conditions? >Any feedback would be appreciated. Not sure if I have a problem or if the >devices are working with in spec. > >Thanks >Scott > > >_______________________________________________ >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/
