Date: Sun, 11 Mar 2001 14:36:12 -0700 From: "Alex Huppenthal" <alex@aspenworks.com> To: <freebsd-isp@FreeBSD.ORG> Subject: ATM and traffic shaping Message-ID: <004101c0aa73$4bdefdc0$1800a8c0@d7k>
next in thread | raw e-mail | index | archive | help
It turns out that traffic shaping for ATM means being able to delay each
cell to a specific egress ('transmit' in english) rate.
If anyone knows how to change all packet sizes to be < 48 bytes (cell header
takes 5), and delay the delivery of each packet to the HARP / Fore card
driver, that would be most miraculous. It would mean that it's possible to
setup Classical IP over ATM on FreeBSD and adjust each PVC by associating an
IP address with it, (unfortuately, just the destination IP address), and
deliver ATM-shaped traffic.
This would then conform to the network's expectation for offered traffic,
and allow us ATM/IP network designers to use FreeBSD in the design.
There's another fix, instead of FreeBSD hosting an ATM port at OC3 rates, it
could host some T1 muxed thing which allows the local interface to throttle
the emitted cells to some rate. This.. would suck.. since an OC3 card is
$800 off the shelf and it plugs and plays nicely on the physical level with
Lucent ATM switches.
No $2000 DS3 CSU, and no $3000 HSSI card required.
If we can't figure out a fix of the FreeBSD Harp stack or add an ioctl to
the 'atm' command that would set the Fore card appropriately per PVC.. we
may be forced (*choke choke* cough-sputter) to use the working Linux ATM
code. I've always wondered what that kernel/environment looked like anyway..
;-)
Cheers!
-Alex
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-isp" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?004101c0aa73$4bdefdc0$1800a8c0>