From: Scott Vermillion (scott_ccie_list@it-ag.com)
Date: Tue Oct 30 2007 - 13:35:38 ART
Hey Simon,
Following a less than inspired workout yesterday, I was relaxing in the hot
tub when a thought occurred to me: traffic egressing subinterface software
queues does not necessarily go directly to the TX ring. There is a physical
interface software queue to be considered as well. But the overall concept
remains the same regardless. So in light of all this, it seems almost
certain (to me) that you must chop up your other traffic as well, as it all
shares some common buffer space, whether software or hardware.
Again, though, my QoS knowledge is pretty shiny and new (and thus subject to
all of the foibles of youth). So it would be interesting to hear from true
QoS gurus (or non-gurus such as myself with an opinion to sport)...
Regards,
Scott
-----Original Message-----
From: Scott Vermillion [mailto:scott_ccie_list@it-ag.com]
Sent: Monday, October 29, 2007 12:14 PM
To: 'Simon Grace'; 'ccielab@groupstudy.com'
Subject: RE: FRTS - One one Subinterface
Well, LOL Simon, I hope you're not talking a load of old, as that's more or
less what I was thinking. But sometimes new knowledge is a little like new
wine...
-----Original Message-----
From: nobody@groupstudy.com [mailto:nobody@groupstudy.com] On Behalf Of
Simon Grace
Sent: Monday, October 29, 2007 11:32 AM
To: Simon Grace; ccielab@groupstudy.com
Subject: RE: FRTS - One one Subinterface
Sorry, I wasn't overly clear...two many can's of coke.
In order to allow VOIP traffic to be interleaved between larger data
conversations ensure the maximum time it takes to transmit a packet
across the Frame Relay network is 10ms.
Now thinking about this (and looking forward to your comments) The only
voice traffic is going to R2, which is via the multipoint subinterface
BUT,
All traffic must go out via the serial 1/4 hardware queue, hence all
traffic even if it's going to the other hub R1 or the point to point
subinterface R5 must be fragmented so as not to jam up the Hardwar queue
Am I thinking correctly?
Or talking a load of old.....
________________________________
From: Simon Grace
Sent: Monday, October 29, 2007 6:25 PM
To: ccielab@groupstudy.com
Subject: FRTS - One one Subinterface
Hi Guys,
Scenario:
R4 has serial interface 1/4
It has two subinterfaces.
The multipoint goes to R1 and R2 with R4 being the HUB.
The point-to-point sub interface goes to R5.
Task is to set up FRTS going to R2 for voice.
Main interfaces is T1, connection between R2 is provisioned at 512kbps.
Remaining bandwidth is to be shared between the connection to R1 (the
other spoke) and R5 the other router on the point-to-point.
We are told that TC must be 10ms and to allow for fragmentation.
So 512000 for the connection to R2 leaves 1024000 divided by 2 = 512000.
So the CIR going to R2, R1 and R5 will be set to 512000.
Now the connection between R4 and R2 is carrying Voice traffic so we are
looking at Fragmentation and Tc of 10ms.
I'm cool with all that!
Now, question, do we need to alter the Tc and have fragmentation going
to R1 (the other spoke) and R5.
I don't think so, but the solution says yes.
Doing a show frame pvc on the pvc going to R2 show everything as it
should be, the two pvc's going to R1 and R5 have the default Tc which I
would expect.
As always, appreciate your comments,
Cheers,
Simon.
This archive was generated by hypermail 2.1.4 : Fri Nov 16 2007 - 13:11:19 ART