iproute2/tc
Eric Dumazet 185d88f99b tc_codel: Controlled Delay AQM
An implementation of CoDel AQM, from Kathleen Nichols and Van Jacobson.

http://queue.acm.org/detail.cfm?id=2209336

This AQM main input is no longer queue size in bytes or packets, but the
delay packets stay in (FIFO) queue.

As we don't have infinite memory, we still can drop packets in enqueue()
in case of massive load, but mean of CoDel is to drop packets in
dequeue(), using a control law based on two simple parameters :

target : target sojourn time (default 5ms)
interval : width of moving time window (default 100ms)

Selected packets are dropped, unless ECN is enabled and packets can get
ECN mark instead.

Usage: tc qdisc ... codel [ limit PACKETS ] [ target TIME ]
                          [ interval TIME ] [ ecn ]

qdisc codel 10: parent 1:1 limit 2000p target 3.0ms interval 60.0ms ecn
 Sent 13347099587 bytes 8815805 pkt (dropped 0, overlimits 0 requeues 0)
 rate 202365Kbit 16708pps backlog 113550b 75p requeues 0
  count 116 lastcount 98 ldelay 4.3ms dropping drop_next 816us
  maxpacket 1514 ecn_mark 84399 drop_overlimit 0

CoDel must be seen as a base module, and should be used keeping in mind
there is still a FIFO queue. So a typical setup will probably need a
hierarchy of several qdiscs and packet classifiers to be able to meet
whatever constraints a user might have.

One possible example would be to use fq_codel, which combines Fair
Queueing and CoDel, in replacement of sfq / sfq_red.

Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Dave Taht <dave.taht@bufferbloat.net>
2012-05-22 14:13:52 -07:00
..
.gitignore Add ignore files to make using git easier 2006-08-08 12:04:38 -07:00
Makefile tc_codel: Controlled Delay AQM 2012-05-22 14:13:52 -07:00
README.last (Logical change 1.3) 2004-04-15 20:56:59 +00:00
em_cmp.c Fix wrong comparison in cmp_print_eopt() 2011-10-07 11:16:15 -07:00
em_meta.c Convert to use rta_getattr_ functions 2012-04-10 08:47:55 -07:00
em_nbyte.c tc: remove dlfcn.h from files that dont need it 2009-11-13 14:14:07 -08:00
em_u32.c tc: remove dlfcn.h from files that dont need it 2009-11-13 14:14:07 -08:00
emp_ematch.l fix build issues with flex ver 2.5 2010-04-22 15:27:42 -07:00
emp_ematch.y ematch: fix warning about yyerror and const 2012-01-03 13:55:00 -08:00
f_basic.c Convert to use rta_getattr_ functions 2012-04-10 08:47:55 -07:00
f_cgroup.c iproute2: Remove unreachable code 2011-07-11 10:13:51 -07:00
f_flow.c Convert to use rta_getattr_ functions 2012-04-10 08:47:55 -07:00
f_fw.c Convert to use rta_getattr_ functions 2012-04-10 08:47:55 -07:00
f_route.c Convert to use rta_getattr_ functions 2012-04-10 08:47:55 -07:00
f_rsvp.c Convert to use rta_getattr_ functions 2012-04-10 08:47:55 -07:00
f_tcindex.c Convert to use rta_getattr_ functions 2012-04-10 08:47:55 -07:00
f_u32.c Convert to use rta_getattr_ functions 2012-04-10 08:47:55 -07:00
m_action.c libnetlink: remove unused junk callback 2011-12-28 10:37:12 -08:00
m_csum.c csum action, fix typo 2012-03-15 14:24:59 -07:00
m_ematch.c Fix NULL pointer reference when using basic match 2010-07-29 18:03:35 -07:00
m_ematch.h ematch related bugfix and cleanup 2008-05-29 11:54:19 -07:00
m_estimator.c Replace "usec" by "time" in function names 2007-03-13 14:42:17 -07:00
m_gact.c Remove trailing whitespace 2006-12-05 10:10:22 -08:00
m_ipt.c Convert to use rta_getattr_ functions 2012-04-10 08:47:55 -07:00
m_mirred.c Remove mirred debug message 2010-03-29 17:32:37 -07:00
m_nat.c tc: remove dlfcn.h from files that dont need it 2009-11-13 14:14:07 -08:00
m_pedit.c Remove trailing whitespace 2006-12-05 10:10:22 -08:00
m_pedit.h Remove trailing whitespace 2006-12-05 10:10:22 -08:00
m_police.c Convert to use rta_getattr_ functions 2012-04-10 08:47:55 -07:00
m_skbedit.c skbedit: fix set-never-used warning 2011-06-29 15:59:02 -07:00
m_xt.c Convert to use rta_getattr_ functions 2012-04-10 08:47:55 -07:00
m_xt_old.c Convert to use rta_getattr_ functions 2012-04-10 08:47:55 -07:00
p_icmp.c Remove trailing whitespace 2006-12-05 10:10:22 -08:00
p_ip.c Remove trailing whitespace 2006-12-05 10:10:22 -08:00
p_tcp.c Remove trailing whitespace 2006-12-05 10:10:22 -08:00
p_udp.c Remove trailing whitespace 2006-12-05 10:10:22 -08:00
q_atm.c Convert to use rta_getattr_ functions 2012-04-10 08:47:55 -07:00
q_cbq.c tc: remove stale code 2010-01-21 10:13:01 -08:00
q_choke.c Convert to use rta_getattr_ functions 2012-04-10 08:47:55 -07:00
q_codel.c tc_codel: Controlled Delay AQM 2012-05-22 14:13:52 -07:00
q_drr.c Convert to use rta_getattr_ functions 2012-04-10 08:47:55 -07:00
q_dsmark.c Convert to use rta_getattr_ functions 2012-04-10 08:47:55 -07:00
q_fifo.c tc: add new queue discipline: head drop fifo 2010-03-03 16:15:44 -08:00
q_gred.c gred: support TCA_GRED_MAX_P attribute 2012-01-20 08:12:24 -08:00
q_hfsc.c HFSC (7) & (8) documentation + assorted changes 2011-11-02 16:33:50 -07:00
q_htb.c tc: remove stale code 2010-01-21 10:13:01 -08:00
q_ingress.c tc: remove stale code 2010-01-21 10:13:01 -08:00
q_mqprio.c iproute2: tc - mqprio formatted print fix 2012-02-22 15:23:12 -08:00
q_multiq.c iproute2: Remove unreachable code 2011-07-11 10:13:51 -07:00
q_netem.c tc-netem: Add support for ECN packet marking 2012-05-22 14:10:21 -07:00
q_prio.c Convert to use rta_getattr_ functions 2012-04-10 08:47:55 -07:00
q_qfq.c Convert to use rta_getattr_ functions 2012-04-10 08:47:55 -07:00
q_red.c Convert to use rta_getattr_ functions 2012-04-10 08:47:55 -07:00
q_rr.c Convert to use rta_getattr_ functions 2012-04-10 08:47:55 -07:00
q_sfb.c tc : SFB flow scheduler 2011-04-12 14:27:37 -07:00
q_sfq.c sfq: add optional RED on top of SFQ 2012-01-20 08:12:22 -08:00
q_tbf.c tc: remove stale code 2010-01-21 10:13:01 -08:00
static-syms.c support static-only systems 2009-11-10 10:44:20 -08:00
tc.c Changing commandline help text to be more uniform... 2009-03-27 11:05:44 -07:00
tc_cbq.c Replace "usec" by "time" in function names 2007-03-13 14:42:17 -07:00
tc_cbq.h (Logical change 1.3) 2004-04-15 20:56:59 +00:00
tc_class.c Convert to use rta_getattr_ functions 2012-04-10 08:47:55 -07:00
tc_common.h add generic size table for qdiscs 2008-09-17 21:57:15 -07:00
tc_core.c add generic size table for qdiscs 2008-09-17 21:57:15 -07:00
tc_core.h add generic size table for qdiscs 2008-09-17 21:57:15 -07:00
tc_estimator.c Introduce TIME_UNITS_PER_SEC to represent internal clock resolution 2007-03-13 14:42:16 -07:00
tc_filter.c Convert to use rta_getattr_ functions 2012-04-10 08:47:55 -07:00
tc_monitor.c update rest to use nl_mgrp 2007-03-13 14:39:05 -07:00
tc_qdisc.c Convert to use rta_getattr_ functions 2012-04-10 08:47:55 -07:00
tc_red.c red: give a hint about burst value 2011-12-01 09:23:43 -08:00
tc_red.h (Logical change 1.3) 2004-04-15 20:56:59 +00:00
tc_stab.c add generic size table for qdiscs 2008-09-17 21:57:15 -07:00
tc_util.c iproute2 - Fix up and simplify variables pointing to install directories 2012-04-12 09:49:10 -07:00
tc_util.h netem: add support for 4 state and GE loss model 2011-12-22 17:08:11 -08:00

README.last

Kernel code and interface.
--------------------------

* Compile time switches

There is only one, but very important, compile time switch.
It is not settable by "make config", but should be selected
manually and after a bit of thinking in <include/net/pkt_sched.h>

PSCHED_CLOCK_SOURCE can take three values:

	PSCHED_GETTIMEOFDAY
	PSCHED_JIFFIES
	PSCHED_CPU


 PSCHED_GETTIMEOFDAY

Default setting is the most conservative PSCHED_GETTIMEOFDAY.
It is very slow both because of weird slowness of do_gettimeofday()
and because it forces code to use unnatural "timeval" format,
where microseconds and seconds fields are separate.
Besides that, it will misbehave, when delays exceed 2 seconds
(f.e. very slow links or classes bounded to small slice of bandwidth)
To resume: as only you will get it working, select correct clock
source and forget about PSCHED_GETTIMEOFDAY forever.


 PSCHED_JIFFIES

Clock is derived from jiffies. On architectures with HZ=100
granularity of this clock is not enough to make reasonable
bindings to real time. However, taking into account Linux
architecture problems, which force us to use artificial
integrated clock in any case, this switch is not so bad
for schduling even on high speed networks, though policing
is not reliable.


 PSCHED_CPU

It is available only for alpha and pentiums with correct
CPU timestamp. It is the fastest way, use it when it is available,
but remember: not all pentiums have this facility, and
a lot of them have clock, broken by APM etc. etc.