HTB vs HFSC, HTB is highly inaccurate

[Denys Fedoryschenko]

Hi again

I test again HTB and HFSC, this time in very simple case.

Regarding HTB - i try without quantum, with different values, without and with
cburst (big and small), with HZ=1000 and default, hysteresis on and off - it
is almost same result.

While i set limit at 57, it can reach 61. HFSC results is perfect.

Here is graph snapshopts.
Loading Image...
You can see points where HTB enabled. Other points if HFSC.

Scripts source is:

HTB http://www.nuclearcat.com/netdev/htb-script.txt
HFSC http://www.nuclearcat.com/netdev/hfsc-script.txt



--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to ***@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html

[Jarek Poplawski]

Post by Denys Fedoryschenko
Hi again
I test again HTB and HFSC, this time in very simple case.
Regarding HTB - i try without quantum, with different values, without and with
cburst (big and small), with HZ=1000 and default, hysteresis on and off - it
is almost same result.
While i set limit at 57, it can reach 61. HFSC results is perfect.
Here is graph snapshopts.
http://www.nuclearcat.com/netdev/htb-problem.png
You can see points where HTB enabled. Other points if HFSC.

Very nice picture, but some tc snapshot stats for qdisc and classes
would be appreciated.

Post by Denys Fedoryschenko
HTB http://www.nuclearcat.com/netdev/htb-script.txt

| echo "class add dev eth0.37 parent 1:1 classid 1:2 htb rate 40Mbit ceil 56Mbit burst 16000b cburst 16000b quantum 16000"
| echo "class add dev eth0.37 parent 1:1 classid 1:3 htb rate 26Mbit ceil 56Mbit burst 16000b cburst 16000b quantum 16000"

I guess you should try with rates 40/16 or 30/26 here.

BTW, it doesn't matter here, but for various reasons it's better to
add classes before filters pointing these classes.

Post by Denys Fedoryschenko
HFSC http://www.nuclearcat.com/netdev/hfsc-script.txt

This link doesn't work for me.

Jarek P.
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to ***@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html

[Denys Fedoryschenko]

Post by Jarek Poplawski
I guess you should try with rates 40/16 or 30/26 here.

Seems i was blind. I forgot main HTB rule - always keep sum or childs rate
less than parent. And i should not listen my boss, cause he dont know
arithmetics. (i ask him about decision, how much for what :-D ).

I will try some experiments now.

But at same time HFSC worked fine.
File fixed.

Post by Jarek Poplawski
BTW, it doesn't matter here, but for various reasons it's better to
add classes before filters pointing these classes.

Post by Denys Fedoryschenko
HFSC http://www.nuclearcat.com/netdev/hfsc-script.txt

This link doesn't work for me.
Jarek P.
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
More majordomo info at http://vger.kernel.org/majordomo-info.html

--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to ***@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html