[65937] in North American Network Operators' Group
RE: Bandwidth Control Question
daemon@ATHENA.MIT.EDU (Dan Ellis)
Fri Dec 19 11:22:09 2003
Date: Fri, 19 Dec 2003 11:21:33 -0500
From: "Dan Ellis" <ellis@corp.ptd.net>
To: "Michel Py" <michel@arneill-py.sacramento.ca.us>,
"Bryan Heitman" <bryan@bryanheitman.com>, <nanog@merit.edu>
Cc: "Claydon, Tom" <Tom.Claydon@DobsonTelco.net>
Errors-To: owner-nanog-outgoing@merit.edu
Thanks. I not very familiar with the integrated DSU T3/E3 command set - =
still used to the good ol HSSI ports. I'll agree, this sounds like a =
better solution if you are using one of the integrated cards.
Yes - you are correct, in my solution both sides should always use a =
traffic-shape or other shaping QoS command. The rate limit is a final =
"police" in case the customer does try to send more than you would like =
them to. Again, agreed, obviously if you can control the port speed =
(above - DSU bandwidth), that's a better solution.
--
Daniel Ellis,=A0CTO, PenTeleData
(610)826-9293
"The only way to predict the future is to invent it."
--Alan Kay
-----Original Message-----
From: Michel Py [mailto:michel@arneill-py.sacramento.ca.us]=20
Sent: Friday, December 19, 2003 10:58 AM
To: Bryan Heitman; nanog@merit.edu
Subject: RE: Bandwidth Control Question
> Bryan Heitman
> Why not simply use configuration option Cisco gives
> you to set your DS3 to 6 meg dsu bandwidth X
That's what I do, works fine.
=A0
> Dan, your suggestion will unncessarily tax his equipment.
Not only that, but the rate-limiting on the input interface will likely =
force the customer to do some QOS at their end too; the discrepancy =
between what the customer's router thinks the bandwidth is and what it =
really is will cause packet loss. I like the solution of the sending =
interface to queue the egress traffic at whatever speed is available =
better.
Michel.
