* Re: [LARTC] hfsc scheduler
2004-10-29 23:06 [LARTC] hfsc scheduler yoyo
@ 2004-10-29 23:20 ` Jason Boxman
2004-10-29 23:39 ` yoyo
` (3 subsequent siblings)
4 siblings, 0 replies; 6+ messages in thread
From: Jason Boxman @ 2004-10-29 23:20 UTC (permalink / raw)
To: lartc
On Friday 29 October 2004 19:06, yoyo wrote:
> hi all,
>
> long time since i posted on the list.
> just wondering if anybody has played around with hfsc and if so could
> he/she share their info on it
http://www-2.cs.cmu.edu/afs/cs.cmu.edu/user/hzhang/WWW/HFSC/
http://www.tik.ee.ethz.ch/~crossbow/rp/plugins/hfsc.html
http://trash.net/~kaber/hfsc/
I'd be curious to see actual examples of usage, though.
--
Jason Boxman
Perl Programmer / *NIX Systems Administrator
Shimberg Center for Affordable Housing | University of Florida
http://edseek.com/ - Linux and FOSS stuff
_______________________________________________
LARTC mailing list / LARTC@mailman.ds9a.nl
http://mailman.ds9a.nl/mailman/listinfo/lartc HOWTO: http://lartc.org/
^ permalink raw reply [flat|nested] 6+ messages in thread* Re: [LARTC] hfsc scheduler
2004-10-29 23:06 [LARTC] hfsc scheduler yoyo
2004-10-29 23:20 ` Jason Boxman
@ 2004-10-29 23:39 ` yoyo
2004-10-30 0:03 ` Andy Furniss
` (2 subsequent siblings)
4 siblings, 0 replies; 6+ messages in thread
From: yoyo @ 2004-10-29 23:39 UTC (permalink / raw)
To: lartc
i remember on this list someone tried hfsc (he had some nice
comparison graphs between hfsc&htb) but i can't seem to find the
message in the archives.. :(
On Fri, 29 Oct 2004 23:33:25 +0000, yoyo <adrian.vasile@gmail.com> wrote:
> me too :)
>
>
>
>
> On Fri, 29 Oct 2004 19:20:35 -0400, Jason Boxman <jasonb@edseek.com> wrote:
> > On Friday 29 October 2004 19:06, yoyo wrote:
> > > hi all,
> > >
> > > long time since i posted on the list.
> > > just wondering if anybody has played around with hfsc and if so could
> > > he/she share their info on it
> >
> > http://www-2.cs.cmu.edu/afs/cs.cmu.edu/user/hzhang/WWW/HFSC/
> > http://www.tik.ee.ethz.ch/~crossbow/rp/plugins/hfsc.html
> > http://trash.net/~kaber/hfsc/
> >
> > I'd be curious to see actual examples of usage, though.
> >
> > --
> >
> > Jason Boxman
> > Perl Programmer / *NIX Systems Administrator
> > Shimberg Center for Affordable Housing | University of Florida
> > http://edseek.com/ - Linux and FOSS stuff
> >
> > _______________________________________________
> > LARTC mailing list / LARTC@mailman.ds9a.nl
> > http://mailman.ds9a.nl/mailman/listinfo/lartc HOWTO: http://lartc.org/
> >
>
> --
> To mess up a Linux box, you need to work at it; to mess up your Windows
> box, you just need to work on it.
>
--
To mess up a Linux box, you need to work at it; to mess up your Windows
box, you just need to work on it.
_______________________________________________
LARTC mailing list / LARTC@mailman.ds9a.nl
http://mailman.ds9a.nl/mailman/listinfo/lartc HOWTO: http://lartc.org/
^ permalink raw reply [flat|nested] 6+ messages in thread* Re: [LARTC] hfsc scheduler
2004-10-29 23:06 [LARTC] hfsc scheduler yoyo
2004-10-29 23:20 ` Jason Boxman
2004-10-29 23:39 ` yoyo
@ 2004-10-30 0:03 ` Andy Furniss
2004-11-01 12:46 ` syrius.ml
2004-11-02 3:18 ` Jason Boxman
4 siblings, 0 replies; 6+ messages in thread
From: Andy Furniss @ 2004-10-30 0:03 UTC (permalink / raw)
To: lartc
yoyo wrote:
> i remember on this list someone tried hfsc (he had some nice
> comparison graphs between hfsc&htb) but i can't seem to find the
> message in the archives.. :(
This one.
Andy.
Vincent Perrier wrote:
> HTB versus HFSC, both qdisc offer the same kind of service,
> if you want to see comparative test results, go to
> http://www.rawsoft.org
> at the line "TEST RESULTS" you will find the results for
> a sharing test and a burst test.
> You will see that both qdisc are good.
Nice comparision, very interesting. Note that you have a small
misconfiguration in your HFSC setup. On page 8 you say "The
shaping is impacted by real time bursts". This is only because
your real-time classes are not part of the link-sharing hierarchy.
If you add link-share curves to the real-time classes which are
equal to the real-time curves shaping won't be impacted.
Regards
Patrick
_______________________________________________
LARTC mailing list / LARTC@mailman.ds9a.nl
http://mailman.ds9a.nl/mailman/listinfo/lartc HOWTO: http://lartc.org/
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [LARTC] hfsc scheduler
2004-10-29 23:06 [LARTC] hfsc scheduler yoyo
` (2 preceding siblings ...)
2004-10-30 0:03 ` Andy Furniss
@ 2004-11-01 12:46 ` syrius.ml
2004-11-02 3:18 ` Jason Boxman
4 siblings, 0 replies; 6+ messages in thread
From: syrius.ml @ 2004-11-01 12:46 UTC (permalink / raw)
To: lartc
I'm using HFSC. But I haven't had time to really understand how to
correctly define curves. So i'm using it the linear way.
I'm just using a rt service curve for my interactive class. (but the
root class is a linear one)
for sure a simple example for a real case could help.
hmm in my case on my 2028/256kbits adsl line: my classes are roughly
defined like this:
root
/ \
interactive bulk_parent
/ \
normal boring_bulk_parent
/ \
syn+small_acks real_boring_bulk
/ | \
host1 host2 host3
(ul rates are near 240kbits, it's a bit less for "boring" classes)
My interactive class is defined like this:
... ls umax ${MTU}b dmax 25ms rate ${MAX}kbit \
rt umax ${MTU}b dmax 25ms rate ${MAX}kbit
25ms, it's a bit less than the latency i have when my link isn't used.
i guess my setup is not really coherent. bulk_parent shouldn't be a
linear service class. And if that's true, all its leaf classes
shouldn't be linear services ones.
another thing, (an example is better than a long long phrase:)
using the same class tree with htb:
host[1-3] upload at 25kB/s, if I start a ftp upload (normal bulk
class), it takes precedence, host[1-3] are slowed and the ftp upload
gets(sends at) more than 25kB/s
I'd like to use HFSC and always have that behavior.
I'm pretty sure a howto-use-hfsc-in-real-context+faq could help. :)
unfortunatly I'm afraid not a lot of people can write it :)
i've read the theory docs, english isn't not my monther tongue language
and for me there's a huge gap between hfsc theory and hfsc in a real
situation.
Anyway, many thanks to Patrick for his work and the documentation he
wrote.
at the moment, for my usage, i prefer to use HFSC rather than HTB.
--
_______________________________________________
LARTC mailing list / LARTC@mailman.ds9a.nl
http://mailman.ds9a.nl/mailman/listinfo/lartc HOWTO: http://lartc.org/
^ permalink raw reply [flat|nested] 6+ messages in thread* Re: [LARTC] hfsc scheduler
2004-10-29 23:06 [LARTC] hfsc scheduler yoyo
` (3 preceding siblings ...)
2004-11-01 12:46 ` syrius.ml
@ 2004-11-02 3:18 ` Jason Boxman
4 siblings, 0 replies; 6+ messages in thread
From: Jason Boxman @ 2004-11-02 3:18 UTC (permalink / raw)
To: lartc
On Monday 01 November 2004 07:46, syrius.ml@no-log.org wrote:
<snip>
> My interactive class is defined like this:
> ... ls umax ${MTU}b dmax 25ms rate ${MAX}kbit \
> rt umax ${MTU}b dmax 25ms rate ${MAX}kbit
Where is the 'ls', 'rt', and the other parameters explained? I'm guessing
'rt' is realtime? What's 'ls'?
<snip>
> I'm pretty sure a howto-use-hfsc-in-real-context+faq could help. :)
> unfortunatly I'm afraid not a lot of people can write it :)
> i've read the theory docs, english isn't not my monther tongue language
> and for me there's a huge gap between hfsc theory and hfsc in a real
> situation.
I think I understand what hfsc is attempting to address, but it's never been
made clear how exactly you interact with Linux's hfsc implementation via the
`tc` binary.
--
Jason Boxman
Perl Programmer / *NIX Systems Administrator
Shimberg Center for Affordable Housing | University of Florida
http://edseek.com/ - Linux and FOSS stuff
_______________________________________________
LARTC mailing list / LARTC@mailman.ds9a.nl
http://mailman.ds9a.nl/mailman/listinfo/lartc HOWTO: http://lartc.org/
^ permalink raw reply [flat|nested] 6+ messages in thread