Have a path like this:
RouterA----uncongested link----RouterB----congested link----RouterC
The congested link is running very full. Users on the far side of
RouterC complain of latency. Ping tests from RouterB to the connected
serial interface on RouterC show a little bit of latency (say, 10-15ms
more than idle on DS3).
At the same time, ping tests from RouterA to the same connected
interface on RouterC show outrageous latency (200-300ms). Have verified
there is not an asymmetric routing issue. Problem is relieved during
non-peak traffic level times.
Question: it seems as though, when pinging from a router across a
directly-connected yet congested link, (RouterB in this case) that the
ping packets seem to automatically get higher priority than other user
traffic transiting the same link.
Is this something I am imagining, or does IOS intentionally queue
locally-originated traffic higher than transit traffic?
The reason this came up is because of user latency complaints on the far
side of RouterC. Traceroutes from RouterC show high latency (from user
perspective/direction) between RouterA and RouterB, but the fact remains
that only the link between RouterB and RouterC is congested. Am trying
to explain why this phenomenon occurs, which seems to misleadingly
displace the actual source of latency from one link to another.
Thank you
This archive was generated by hypermail 2b29 : Sun Aug 04 2002 - 04:13:10 EDT