North American Network Operators Group

Date Prev | Date Next | Date Index | Thread Index | Author Index | Historical

Re: long delay between circuits..

  • From: jhyang
  • Date: Wed Sep 23 11:57:19 1998

I got several replies,, and I investigated it more.

We are using tiered T3 service from uu.net, and I suspect that
uu.net use ATM QoS scheme between hop 4 and hop 5.
(We are using 30 Mbps from uu.net)

(But, I knew that the way of tiered bandwidth was using control
 of CSU -LARSCOM-.)

Is it true that uu.net use ATM QoS scheme (i.e. rate control 
by leaky bucket algorithm) for shaping bandwidth ?

When the link has avaible bandwidth, the response time of PING
is much faster. But the RTT of PING is slow when the link goes
to peak rate. --But there are no *packet loss*--

I think it is similar to *rate control*.

Is there any comments ?

---Jaeho Yang.

Harold Willison wrote:
> 
> At 03:49 PM 9/16/98 +0900, you wrote:
> >Hello,
> >
> >I want to hear advices from gurus,,
> >
> > Our network use transit service from UU.NET in PaloAlto IX.
> >
> > ## the result of traceroute from our network to www.sun.com
> > traceroute to Sun.COM (192.9.49.33), 30 hops max, 40 byte packets
> > --<snip>--
> > 4 INET-PA-GW.nuri.net (203.235.119.253) 147 ms 145 ms 143 ms
> > 5 911.Hssi5-0.GW1.PAO1.ALTER.NET (157.130.192.145) 170 ms 169 ms
> >173 ms <[email protected]@@@
> > 6 119.ATM3-0.XR1.SCL1.ALTER.NET (146.188.144.78) 486 ms 434 ms 441
> >ms <[email protected]@@@
> > 7 146.188.145.157 (146.188.145.157) 432 ms 440 ms 443 ms
> > 8 core7-hssi5-0.SanFrancisco.mci.net (206.157.77.73) 408 ms 457 ms
> >425 ms
> > --<snip>--
> 
> I would suggest getting UUnet to do perform 2 traceroutes back to your originating address.
> 
> The problem could be that when your traffic hits hop 6 it has a different return route which
> has poorer performance.
> If UUnet performs the traceroutes from the hop 5 and hop 6 routers towards you you will see the path both
> takes. If they are both the same path then it would definately appear to be a problem on their
> link. If the route back from hop 5 is different than from hop 4 I would suggest checking for the problem spot
> in that traceroute.
> 
> *-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*
> | Harold Willison AGIS Network Engineering |
> * Senior Network Engineer 313-730-5151 *
> | [email protected] 313-730-1130 x-5649 |
> | [email protected] 24 hours a day, 7 days a week |
> | http://www.agis.net |
> \*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*/