LARTC - Jul 2002 - Anything out there that is similar to Cisco''s WFQ?

> From: "CIT/Paul" <xerox@foonet.net>
 > Any help would be greatly appreciated :) This is much better than SFQ
:>

Sounds like SFQ to me.  Can you tell us what the differences are?

Nooooo SFQ is not like WFQ... WRR is the closest thing to cisco''s
fair-queue..
WRR keeps track of the connections using the ip_conntrack .. that''s
sort of
what
cisco''s fair-queue does and it checks the bandwidth streams and gives
lower
priority
to the higher streams and larger packets.. it''s meant to reduce latency
for
traffic
shaping and it does :)
I haven''t tried WRR but it looks like the closest thing to it although
it
doesn''t
take everything in to account as cisco''s flow based WFQ does..

Paul

-----Original Message-----
From: Don Cohen [mailto:don-lartc@isis.cs3-inc.com]
Sent: Wednesday, July 10, 2002 3:55 PM
To: xerox@foonet.net; lartc@mailman.ds9a.nl
Subject: Anything out there that is similar to Cisco''s WFQ?


 > From: "CIT/Paul" <xerox@foonet.net>
 > Any help would be greatly appreciated :) This is much better than SFQ
:>

Sounds like SFQ to me.  Can you tell us what the differences are?

Paul writes:
 > Nooooo SFQ is not like WFQ... WRR is the closest thing to cisco''s
 > fair-queue..
 > WRR keeps track of the connections using the ip_conntrack ..
that''s sort of
 > what
 > cisco''s fair-queue does and it checks the bandwidth streams and
gives lower
 > priority
 > to the higher streams and larger packets.. it''s meant to reduce
latency for
 > traffic
 > shaping and it does :)
 > I haven''t tried WRR but it looks like the closest thing to it
although it
 > doesn''t
 > take everything in to account as cisco''s flow based WFQ does..

This is not very convincing.  Do you actually know how WFQ
works?  If so, please tell us.  The doc you sent did not describe how
it works but what the effects are, and those are entirely consistent
with what SFQ does. 
High bandwidth flows are limited, low bandwidth flows get lower
latency.  Can you describe some effect that''s different?

Don Cohen wrote:
> > From: "CIT/Paul" <xerox@foonet.net>
> > Any help would be greatly appreciated :) This is much better than SFQ
:>
>
>Sounds like SFQ to me.  Can you tell us what the differences are?
>  
>
PRIO''d SFQ.

If you had classful PRIO with SFQ on each band, you''d probably have a 
similar effect to what''s been described; just a guess though.  It seems
the desire is to ''ignore'' low-priority bands if high-priority
bands have
traffic, and to balance between those sessions.
-- 
Michael T. Babcock
CTO, FibreSpeed Ltd.

=?iso-8859-1?Q?John_B=E4ckstrand?= writes:
 > > This is not very convincing.  Do you actually know
 > how WFQ
 > > works?  If so, please tell us.  The doc you sent did
 > not describe how
 > > it works but what the effects are, and those are
 > entirely consistent
 > > with what SFQ does.
 > > High bandwidth flows are limited, low bandwidth flows
 > get lower
 > > latency.  Can you describe some effect that''s
 > different?
 > 
 > I read a bit on WFQ earlier, Im not grasping it totally
 > and I dont know every implementation detail, but I
 > think its basically WRR but taking actual bandwidth
 > usage into account, and not just packet-counts. Well,
 > try this:
 > 
 > http://www.sics.se/~ianm/WFQ/wfq_descrip/node21.html

This sounds just like SFQ except for the weights.
I have a variant of SFQ that does support weights if that''s important.
It''s easy to add.  (The hard part is the code that allows you to
configure the weights.)

> This is not very convincing.  Do you actually know
how WFQ
> works?  If so, please tell us.  The doc you sent did
not describe how
> it works but what the effects are, and those are
entirely consistent
> with what SFQ does.
> High bandwidth flows are limited, low bandwidth flows
get lower
> latency.  Can you describe some effect that''s
different?

I read a bit on WFQ earlier, Im not grasping it totally
and I dont know every implementation detail, but I
think its basically WRR but taking actual bandwidth
usage into account, and not just packet-counts. Well,
try this:

http://www.sics.se/~ianm/WFQ/wfq_descrip/node21.html

Im sure you all can get more out of it than me, a total
newbie to queueing theory and QoS.

---
John Bäckstrand

> =?iso-8859-1?Q?John_B=E4ckstrand?= writes:
>  > > This is not very convincing.  Do you actually
know
>  > how WFQ
>  > > works?  If so, please tell us.  The doc you sent
did
>  > not describe how
>  > > it works but what the effects are, and those are
>  > entirely consistent
>  > > with what SFQ does.
>  > > High bandwidth flows are limited, low bandwidth
flows
>  > get lower
>  > > latency.  Can you describe some effect that''s
>  > different?
>  >
>  > I read a bit on WFQ earlier, Im not grasping it
totally
>  > and I dont know every implementation detail, but I
>  > think its basically WRR but taking actual
bandwidth
>  > usage into account, and not just packet-counts.
Well,
>  > try this:
>  >
>  >
http://www.sics.se/~ianm/WFQ/wfq_descrip/node21.html
>
> This sounds just like SFQ except for the weights.
> I have a variant of SFQ that does support weights if
that''s important.
> It''s easy to add.  (The hard part is the code that
allows you to
> configure the weights.)
I was under the impression that the weights of WFQ isnt
actually supposed to be set manually, but rather
automatically. This page has a nice picture of WFQ (I
think)

http://www.cisco.com/warp/public/732/Tech/wfq/

It says: "Weight determined by:
*Required QoS (IP Procedure, RSVP)
*Flow throughput inversely proportional
*Frame relay FECN, BECN, DE (for FR Traffic)"

Only think I actually understood was "Flow throughput
inversely proportional" which is a property I am
looking for when trying to find a traffic control
implementation.

---
John Bäckstrand

Btw, about the original question (havent got the
original email left), there is a WFQ implementation for
ALTQ and FreeBSD, but it seems to not work too well:

http://www.criticalsoftware.com/research/pdf/Paper-PS.p
df

http://corn.eos.nasa.gov/qos/qos_results_summary_july98
.html

---
John Bäckstrand