[c-nsp] Weird Ping Response Times

Peter Rathlev peter at rathlev.dk
Sun Oct 10 07:50:29 EDT 2010


On Sat, 2010-10-09 at 20:35 -0400, Dominic wrote:
> My voice SBC (Acme  Packets) shares the same subnet, and even the same
> Cisco switch, with a  couple of other devices (including a Cisco GSR
> 12800, Cisco Pix, and a Cisco 7206VXR). When pinging the SBC from
> non-cisco devices, the response time is 0/0/0 ms, as one would expect.
> When pinging from the cisco device to any other device on the same
> switch, other than the SBC, the response time is also 0/0/0 ms.

I would never expect "0 ms" RTT, unless you're truncating the
result. :-)

> However, when pinginh from any of the cisco devices -PIX, GSR, 7206-
> to the SBC, for some reason, the response time is an alarming  1/24/40
> ms. This ping response time -from cisco devices on the same subnet
> directed connected to the same switch -is worse than the response time
> to devices that are on totally different networks!
> 
> Does anybody have any idea what could wrong, or what I should be
> looking to adjust?

There isn't necessarily anything wrong. Keep in mind what you're
actually testing with your ping probe:

 Host A "ping" process creates packet, hands it over to the transmit
 mechanism of the device. Packet traverses network and lands at host B.
 The receive mechanism takes the packet and hands it over to a specific
 "echo reply" process. This process composes a reply packet and a mirror
 copy of the process starts.

So you're really testing a lot of things not related to forwarding
packets. Creating the request and reply packets will take much time for
devices not designed for that specific task. Network equipment is
normally designed to forward packets, not reply to ICMP Echo Requests.

If you want to find out what the forwarding delay is, setup dedicated
measuring devices, e.g. a pair of ISRs with RTR/SAA/IP SLA configured.

-- 
Peter




More information about the cisco-nsp mailing list