* QoS in Netfilter
@ 2014-04-16 8:57 Nikolai Lusan
0 siblings, 0 replies; 2+ messages in thread
From: Nikolai Lusan @ 2014-04-16 8:57 UTC (permalink / raw)
To: netfilter
[-- Attachment #1: Type: text/plain, Size: 140 bytes --]
Hi,
So the merits of MARK vs. CLASSIFY? What do people think is the best way
of using QoS?
--
Nikolai Lusan <nikolai@lusan.id.au>
[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 819 bytes --]
^ permalink raw reply [flat|nested] 2+ messages in thread
* Re: QoS in Netfilter
@ 2014-04-18 1:08 Neal Murphy
0 siblings, 0 replies; 2+ messages in thread
From: Neal Murphy @ 2014-04-18 1:08 UTC (permalink / raw)
To: netfilter
On Wednesday, April 16, 2014 04:57:43 AM Nikolai Lusan wrote:
> Hi,
>
> So the merits of MARK vs. CLASSIFY? What do people think is the best way
> of using QoS?
In a short sentence: Use CONNMARK to identify types of traffic (even when
using a higher-layer sniffer); use CLASSIFY to assign types of traffic to
classes. That's really the best you can hope to do until a simple and well-
supported layer 7 analyzer is available. And don't even think about trying to
classify encrypted streams.
Alas, Traffic Control needs a complete rewrite to handle modern traffic flows.
HTB is nice. And works very well once it's understood. But it's far too
complicated and doesn't really handle today's net usages very well. I've an
idea what to do, but don't have the time to experiment. Think amalgamation of
HTB and conntrack.
^ permalink raw reply [flat|nested] 2+ messages in thread
end of thread, other threads:[~2014-04-18 1:08 UTC | newest]
Thread overview: 2+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2014-04-16 8:57 QoS in Netfilter Nikolai Lusan
-- strict thread matches above, loose matches on Subject: below --
2014-04-18 1:08 Neal Murphy
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).