Linux Advanced Routing and Traffic Control list
 help / color / mirror / Atom feed
From: George Alexandru Dragoi <waruiinu@gmail.com>
To: lartc@vger.kernel.org
Subject: [LARTC] HFSC again.
Date: Wed, 22 Dec 2004 14:17:35 +0000	[thread overview]
Message-ID: <3063e504122206173a65963f@mail.gmail.com> (raw)

       Hello list,

I am now using hfsc for QOS. I have something like this:

tc qdisc add dev ethx root handle 1: hfsc default 9
tc class add dev ethx parent 1: classid 1:1 hfsc ls m2 70mbit ul m2 70mbit

tc class add dev ethx parent 1:1 classid 1:9 hfsc ls m2 7mbit

#DNS stuff follows
tc class add dev ethx parent 1:1 classid 1:2 hfsc rt m1 12mbit d 1ms m2 1mbit

#Half-life
tc class add dev ethx parent 1:1 classid 1:3 hfsc rt m1 24mbit d 1ms m2 2mbit

#Some packets with important flags like SYN SYN,ACK small ACKs
tc class add dev ethx parent 1:1 classid 1:4 hfsc rt m1 12mbit d 1ms m2 1mbit

#www traffic
tc class add dev ethx parent 1:1 classid 1:5 hfsc ls m2 20mbit

#for every client i do this
tc class add dev ethx parent 1:5 classid 1:$wwwclient hfsc rt m2
128kbit ls m2 128kbit
.... And so on for everybody...

#bulk traffic
tc class add dev ethx parent 1:1 classid 1:6 hfsc ls m2 30mbit

#for every client i have ...
tc class add dev ethx parent 1:6 classid 1:$bulkclient ls m2 128kbit
... so on for everybody...


This setup works ok, but unfortunately with not enough knowledge about
what i am doing. The rt classes are meant to guarantee a bandwidth at
an rate (m2). I also used m1 and d parameter, but without thinking
what they really do. After reading
http://www.tik.ee.ethz.ch/~crossbow/rp/plugins/hfsc.html I can say m1
and d has nothing to do with m2 (m1 is calculated based on that
formula), unlike the way i did that, using d as 1ms and m1 as 12 times
greater than m2. After saying those, i have some questions, hopefully
the'll clarify most of my wanders regarding HFSC:

1. How m1 and d really behave, should they be considered and
calculated based on the formula from the link above? Or is just a way
to define how the speed increase or decrease from m1 to m2 and
viceversa into a periodic function (with d as the T).
2. How m1 and d behave when used in different service courves types
(ls, rt and ul).
3. If i use both rt and ls parametters, is ls used for the remained
bandwidth? (i guess Yes)
4. Is it ok to consider the default class as i would do with HTB? I
mean, i usually make the default class in HTB very small and if i see
it does pretty much traffic, then there is something wrong. In HFSC
many exemples gives 10% of bandwidth to default class.
5. In my setup, does classes from bulk traffic will ever get the
maximum of the "most" parent class (70mbit)?

Thanks in advance.
_______________________________________________
LARTC mailing list / LARTC@mailman.ds9a.nl
http://mailman.ds9a.nl/mailman/listinfo/lartc HOWTO: http://lartc.org/

                 reply	other threads:[~2004-12-22 14:17 UTC|newest]

Thread overview: [no followups] expand[flat|nested]  mbox.gz  Atom feed

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=3063e504122206173a65963f@mail.gmail.com \
    --to=waruiinu@gmail.com \
    --cc=lartc@vger.kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox