From: Scott Morris (swm@emanon.com)
Date: Tue Jan 03 2006 - 16:13:26 GMT-3
It's not so much that you can send one packet every 10ms because that often
depends on the size of your packets or fragmentation scheme. But if you
have adjusted the fragmentation to go along with this (on slow speed links)
and have appropriate queuing strategy in place, the concept is that you are
setting up a delay budget. And since you measure every 10ms, you know what
your anticipated budget is on that link.
As Leigh pointed out, the maximum one-way delay in voice should be no more
than 150ms. So, depending on how many links and what speed they may be in
your network, you have a lot of decisions to make!
The voice sample size happens to be 10ms. However, default mechanism of
sending VoIP is two samples per packet (50 packets per second). So you MAY
take that into your calculation too.
But be aware that most of this stuff is calculated as an "engineered" first
starting point, then experience may make you tweak it further. That's a
fancy way of saying that your mileage may vary!
You aren't setting a fixed inter-packet delay, but more of a maximum one
when it comes to interleaving your important (voice) stuff in with
everything else that you may be running.
HTH,
Scott
-----Original Message-----
From: nobody@groupstudy.com [mailto:nobody@groupstudy.com] On Behalf Of
CCIEin2006
Sent: Tuesday, January 03, 2006 1:20 PM
To: Bob Sinclair
Cc: Leigh Harrison; Cisco certification
Subject: Re: FRTS Time Interval
I almost got it now, just need to clear some things up. Please correct me
if I'm wrong:
1. The point of FRTS is so that we create this artificial congestion so that
we decide which packets are dropped rather than let the carrier randomly
drop packets?
2. Lets say our CIR allowed us to send 1 voice packet every 10 ms. If we
used a 10 ms interval we would send exactly 1 packet every 10 ms. But if we
used a 100ms interval, we can possible send all 10 packets in the first 50
ms, then wait 50ms before sending more packets?
3. Basically is this technique a way to create a fixed inter-packet delay?
On 1/3/06, Bob Sinclair <bob@bobsinclair.net> wrote:
>
> CCIEin2006,
>
> I believe the assumption behind the scenario is that we need to do
> FRTS to avoid potential policing or egress blocking. In other words,
> we cannot utilize all the potential bandwidth on the physical
> interface, so we use FRTS to create a kind of artificial congestion.
>
> It would be necessary in this scenario to create a priority queue for
> the voice traffic so that it does not experience queuing delay. But
> we are still only sending bytes to the interface, by default, every
> 125 ms. The voice packet may be "first in the boat", but still the
> boat is too infrequent, so we still need to decrease the Tc to
> something that does not blow the end-to-end delay budget.
>
> But sure, for voice traffic you would like to avoid FRTS entirely if
> you can.
>
> HTH,
>
>
> Bob Sinclair
> CCIE #10427, CCSI 30427
> www.netmasterclass.net
>
> ----- Original Message -----
> From: CCIEin2006
> To: Leigh Harrison
> Cc: Cisco certification
> Sent: Tuesday, January 03, 2006 12:02 PM
> Subject: Re: FRTS Time Interval
>
>
> Why does it have to wait a full 125ms for the boat? If there is enough
> bandwidth the packet should be sent immediatley, no? If there is
> congestion how is playing with time intervals going to give you extra
> bandwidth?
>
> On 1/3/06, Leigh Harrison <ccileigh@gmail.com> wrote:
> >
> > Hey there,
> >
> > You have to think of it in terms of overall delay. Cisco recommends
> > between 150ms and 200ms for an end to end VoIP conversation to be of
> > good quality.
> >
> > If your Tc is set to 125ms and a voice packet misses the boat for
> > that time period, then he's got to wait 125ms before the next boat
arrives.
> > This takes a big chunk out of your total budget of 150ms (for
> > example), leaving you with only 25ms to play with.
> >
> > If your Tc is set to 10ms, then our brave stranded packet will only
> > have to wait for 10ms for the next boat to arrive, taking much less
> > of a bite out of our 150ms budget. Which is a good thing.
> >
> > LH
> >
> >
> >
> >
> > CCIEin2006 wrote:
> >
> > >Can someone please explain how changing the default Tc from 125ms
> > >to
> 10ms
> > >benefits you when running voice traffic over a frame relay link?
> > >I understand that voice traffic is very intolerant of delay and
> > >jitter
> > but
> > >how does changing your time interval help in this situation? Do the
> > packets
> > >get broken up as in frame relay fragmentation?
> > >
> > >Maybe someone can explain what role the time interval plays in
> > >traffic shaping?
> > >
> > >___________________________________________________________________
> > >____ Subscription information may be found at:
> > >http://www.groupstudy.com/list/CCIELab.html
>
> ______________________________________________________________________
> _ Subscription information may be found at:
> http://www.groupstudy.com/list/CCIELab.html
>
> ______________________________________________________________________
> _ Subscription information may be found at:
> http://www.groupstudy.com/list/CCIELab.html
This archive was generated by hypermail 2.1.4 : Wed Feb 01 2006 - 07:45:47 GMT-3