From: amilabs (amilabs@optonline.net)
Date: Tue Dec 10 2002 - 14:05:54 GMT-3
Try to compare the timestamps before and after queuing. Try to keep your
tests the same, no of packets etc, so as to not skew your results..
Agilent Advisor head:)))
-----Original Message-----
From: nobody@groupstudy.com [mailto:nobody@groupstudy.com]On Behalf Of
Chuck Church
Sent: Tuesday, December 10, 2002 11:22 AM
To: Brennan_Murphy@NAI.com; ghie_pogi@yahoo.com; ccielab@groupstudy.com
Subject: Re: Priority Queueing
Sorry, I meant that Sniffer wouldn't tell you much in this case, because
there's no congestion. But it is the only way to see the actual order of
packets being sent out. It's easy to analyze with a slow, congested line
and TCP, because you should see a data packet leave the source right after
the destination sends an acknowledgement. Turn off the queuing, and you'll
see bigger gaps between data and acks. Wow, you NAI guys are defensive :^)
Great product though.
Thanks,
Chuck Church
CCIE #8776, MCNE, MCSE
----- Original Message -----
From: <Brennan_Murphy@NAI.com>
To: <cchurch@optonline.net>; <ghie_pogi@yahoo.com>; <ccielab@groupstudy.com>
Sent: Tuesday, December 10, 2002 11:01 AM
Subject: RE: Priority Queueing
> I agree with Chuck that this scenario is not ripe for
> seeing the potential effects of QoS.
>
> However, I wanted to add a small comment concerning Sniffer
> and similar tools: if you have a strong definition of
> what it means to verify QoS, there is no better tool.
> If you have a loose version....eh, lots of ways to
> do that. Pick any flavor of QoS and if you really
> want to visualize and understand it's impact, a protocol
> sniffer is the only way to truly get there.
>
>
>
> -----Original Message-----
> From: Chuck Church
> To: Angelo De Guzman; ccielab@groupstudy.com
> Sent: 12/10/02 7:24 AM
> Subject: Re: Priority Queueing
>
> Angelo,
>
> The queueing methods always work with outbound traffic. Typically
> used
> with slower interfaces though. Queueing is always working, but like you
> said, if the traffic load isn't high, it's really not giving you any
> benefit. I don't think you'll see anything useful with Sniffer. The
> router
> can send the traffic pretty much as soon as it arrives at the outgoing
> FE
> interface. The queue never fills beyond a couple packets, so no
> prioritization ever takes effect. I think you've just got to trust the
> 'sh
> queuexxx' commands. Or else set up back-to-back aux ports, and create a
> very slow link. Then you'll see the difference, especially if you run
> an
> FTP session, and prioritize telnet. That's where it shines.
>
> Chuck Church
> CCIE #8776, MCNE, MCSE
>
>
> ----- Original Message -----
> From: "Angelo De Guzman" <ghie_pogi@yahoo.com>
> To: <ccielab@groupstudy.com>
> Sent: Tuesday, December 10, 2002 2:34 AM
> Subject: Priority Queueing
>
>
> > HI To All,
> > Wanted to test PQ on my FE interface? I want to
> > prioritize FTP. I have a sniffer to see if it is
> > working. My questions are the following:
> > 1. PQ is applied in what direction?
> > 2. Will my queueing work if my bandwidth utilization
> > is just 1% ? I remember reading that queueing will
> > only work if the utilization is very high.
> >
> > I'm thinking of putting a traffic generator to
> > inject other types of traffic to see if my PQ will
> > work. But I don't have a traffic generator. 8-(
> >
> > Thanks In Advance,
> > Angelo De Guzman
> >
> >
> >
> > __________________________________________________
> > Do you Yahoo!?
> > Yahoo! Mail Plus - Powerful. Affordable. Sign up now.
> > http://mailplus.yahoo.com
> > .
> .
.
.
This archive was generated by hypermail 2.1.4 : Fri Jan 17 2003 - 17:21:42 GMT-3