North American Network Operators Group

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

RE: TCP congestion

  • From: Brian Knoll \(TTNET\)
  • Date: Thu Jul 12 17:00:42 2007

Are you using TCP offloading on your windows box?  I have seen issues
with that in the past where it was dropping data.  Turn it off and see
if the issue goes away.  

Are other the other connections traversing this path seeing the same

Still - the only definitive way to solve the problem is by getting
captures from both ends.  If you can isolate your wan with taps on each
side and see packets being dropped, you know it's your ATM circuit.  

QOS will not help you if you aren't exceeding bandwidth.  

Brian Knoll
Senior Network Engineer, TTNET
312-698-6017 desk
312-823-0957 mobile

-----Original Message-----
From: [email protected] [mailto:[email protected]] On Behalf Of
Philip Lavine
Sent: Thursday, July 12, 2007 3:28 PM
To: Stephen Wilcox
Cc: nanog
Subject: Re: TCP congestion

I just don't understand how if there is 1 segment that gets lost how
this could translate to such a catastrophic long period of slow-start.
How can I minimize the impact of  the inevitable segment loss/out of
order over a WAN. Is QoS the only option?

----- Original Message ----
From: Stephen Wilcox <[email protected]>
To: Philip Lavine <[email protected]>
Cc: nanog <[email protected]>
Sent: Thursday, July 12, 2007 1:09:24 PM
Subject: Re: TCP congestion

Well, if its out of order its the same as if its lost or delayed, it
needs to see that missing segment before the window is full

As mentioned you need to get dumps from both ends, you will almost
definitely find that you have packet loss which tripped tcp's slow start


On Thu, Jul 12, 2007 at 12:02:49PM -0700, Philip Lavine wrote:
> Even if the segment was received out of order what would cause
congestion avoidance to starve the connection of legitimate traffic for
15 to 20 seconds? That is the core of the problem.
> ----- Original Message ----
> From: Fred Baker <[email protected]>
> To: Brian Knoll <[email protected]>
> Cc: Philip Lavine <[email protected]>; nanog <[email protected]>
> Sent: Thursday, July 12, 2007 11:56:06 AM
> Subject: Re: TCP congestion
> On Jul 12, 2007, at 11:42 AM, Brian Knoll ((TTNET)) wrote:
> > If the receiver is sending a DUP ACK, then the sender either never
> > received the first ACK or it didn't receive it within the timeframe
> > expected.
> or received it out of order.
> Yes, a tcpdump trace is the first step.
> Be a better Globetrotter. Get better travel answers from someone who
knows. Yahoo! Answers - Check it out.

Pinpoint customers who are looking for what you sell.