High roundtrips inside the backbone

[asanes]

Hello, I am having problems with one of the routers of the backbone. To illustrate the probelm let me briefly describe the topology. There are 5 routers connected to a catalyst 5000 switches in the same vlan, one router is a GSR, another router has a fastethernet portchannel, other has a gigabit ethernet conenction and the two remaining are connected via fastethernet. The problem i am having is with one of the fastethernet connected router, by the way the cable this router use to connect to the switch is a STP cable, this router handle about 64 Mbps of traffic in high peak hours. The problem i am having is when pinging any other router inside the vlan,the round trip is of about 400 mseg or more. The situation change from time to time and that change can occur when having almost the same amount of traffic through the interface. There is also a lot of packets loss out of the interface of the switch where this router is connected to. Plus, this problem is not happen to the other fastethernet connected router. the parameters of cpu in the problematic router are OK. The buses in the switch is OK too.

I appreciate any comment or recommendations about this topic,,

[steve.barlow]

When a packet destination is the router itself, this packet has to be process-switched. The processor has to handle the information from this packet and send an answer back. This is not the main goal of a router. By definition, a router is built to route packets. Answering a ping is offered as a best-effort service. To test if the router is slow, ping through the router, then it will be fast switched/cef/etc and will give you an idea of normal traffic. What is the routers CPU levels? Can you put a sniffer on to see how long the router holds the packet? Also check why there is packet loss on the switch port and look at buffers/queue on router. Check arp and cam tables. Check path the packets take on your core switches, how many hops, look at those switches?

Hope it helps, let us know, sounds interesting.

Steve

[asanes]

Hello Steve,

I want to tell you about how the situation is going. After looking at the router and switches on the backbone we concluded that the problem probably was that distributed cef was disable on the router, this because the vips on the router (vip 2-40 32 MB) couldn´t handle all the traffic we have on that router and dcef, so we decided to upgrade the vips in some slots, but we still couldn´t run dcef cause we didn´t have enough vips for all slots. So, cause this router is at the distribution level and was connected to the routers vlan through fast ethernet while other routers are connected via giga or portchannel, we decided to configure a portchannel on the routers vlan of the router with two fastethernet in differents vips in order to ease confestion on that port.

Thanks for your interest, and we´ll keep in contact trough this connection.