Blame man/man8/tc-stab.8

Packit d3f73b
.TH STAB 8 "31 October 2011" iproute2 Linux
Packit d3f73b
.
Packit d3f73b
.SH NAME
Packit d3f73b
tc\-stab \- Generic size table manipulations
Packit d3f73b
.
Packit d3f73b
.SH SYNOPSIS
Packit d3f73b
.nf
Packit d3f73b
tc qdisc add ... stab
Packit d3f73b
.RS 4
Packit d3f73b
[ \fBmtu\fR BYTES ] [ \fBtsize\fR SLOTS ]
Packit d3f73b
[ \fBmpu\fR BYTES ] [ \fBoverhead\fR BYTES ]
Packit d3f73b
[ \fBlinklayer\fR { adsl | atm | ethernet } ] ...
Packit d3f73b
.RE
Packit d3f73b
.fi
Packit d3f73b
Packit d3f73b
.SH OPTIONS
Packit d3f73b
For the description of BYTES \- please refer to the \fBUNITS\fR
Packit d3f73b
section of \fBtc\fR(8).
Packit d3f73b
Packit d3f73b
.IP \fBmtu\fR 4
Packit d3f73b
.br
Packit d3f73b
maximum packet size we create size table for, assumed 2048 if not specified explicitly
Packit d3f73b
.IP \fBtsize\fR
Packit d3f73b
.br
Packit d3f73b
required table size, assumed 512 if not specified explicitly
Packit d3f73b
.IP \fBmpu\fR
Packit d3f73b
.br
Packit d3f73b
minimum packet size used in computations
Packit d3f73b
.IP \fBoverhead\fR
Packit d3f73b
.br
Packit d3f73b
per\-packet size overhead (can be negative) used in computations
Packit d3f73b
.IP \fBlinklayer\fR
Packit d3f73b
.br
Packit d3f73b
required linklayer specification.
Packit d3f73b
.PP
Packit d3f73b
.
Packit d3f73b
.SH DESCRIPTION
Packit d3f73b
.
Packit d3f73b
Size tables allow manipulation of packet sizes, as seen by the whole scheduler
Packit d3f73b
framework (of course, the actual packet size remains the same). Adjusted packet
Packit d3f73b
size is calculated only once \- when a qdisc enqueues the packet. Initial root
Packit d3f73b
enqueue initializes it to the real packet's size.
Packit d3f73b
Packit d3f73b
Each qdisc can use a different size table, but the adjusted size is stored in
Packit d3f73b
an area shared by whole qdisc hierarchy attached to the interface. The effect is
Packit d3f73b
that if you have such a setup, the last qdisc with a stab in a chain "wins". For
Packit d3f73b
example, consider HFSC with simple pfifo attached to one of its leaf classes.
Packit d3f73b
If that pfifo qdisc has stab defined, it will override lengths calculated
Packit d3f73b
during HFSC's enqueue; and in turn, whenever HFSC tries to dequeue a packet, it
Packit d3f73b
will use a potentially invalid size in its calculations. Normal setups will
Packit d3f73b
usually include stab defined only on root qdisc, but further overriding gives
Packit d3f73b
extra flexibility for less usual setups.
Packit d3f73b
Packit d3f73b
The initial size table is calculated by \fBtc\fR tool using \fBmtu\fR and
Packit d3f73b
\fBtsize\fR parameters. The algorithm sets each slot's size to the smallest
Packit d3f73b
power of 2 value, so the whole \fBmtu\fR is covered by the size table. Neither
Packit d3f73b
\fBtsize\fR, nor \fBmtu\fR have to be power of 2 value, so the size
Packit d3f73b
table will usually support more than is required by \fBmtu\fR.
Packit d3f73b
Packit d3f73b
For example, with \fBmtu\fR\~=\~1500 and \fBtsize\fR\~=\~128, a table with 128
Packit d3f73b
slots will be created, where slot 0 will correspond to sizes 0\-16, slot 1 to
Packit d3f73b
17\~\-\~32, \&..., slot 127 to 2033\~\-\~2048. Sizes assigned to each slot
Packit d3f73b
depend on \fBlinklayer\fR parameter.
Packit d3f73b
Packit d3f73b
Stab calculation is also safe for an unusual case, when a size assigned to a
Packit d3f73b
slot would be larger than 2^16\-1 (you will lose the accuracy though).
Packit d3f73b
Packit d3f73b
During the kernel part of packet size adjustment, \fBoverhead\fR will be added
Packit d3f73b
to original size, and then slot will be calculated. If the size would cause
Packit d3f73b
overflow, more than 1 slot will be used to get the final size. This of course
Packit d3f73b
will affect accuracy, but it's only a guard against unusual situations.
Packit d3f73b
Packit d3f73b
Currently there are two methods of creating values stored in the size table \-
Packit d3f73b
ethernet and atm (adsl):
Packit d3f73b
Packit d3f73b
.IP ethernet 4
Packit d3f73b
.br
Packit d3f73b
This is basically 1\-1 mapping, so following our example from above
Packit d3f73b
(disregarding \fBmpu\fR for a moment) slot 0 would have 8, slot 1 would have 16
Packit d3f73b
and so on, up to slot 127 with 2048. Note, that \fBmpu\fR\~>\~0 must be
Packit d3f73b
specified, and slots that would get less than specified by \fBmpu\fR will get
Packit d3f73b
\fBmpu\fR instead. If you don't specify \fBmpu\fR, the size table will not be
Packit d3f73b
created at all (it wouldn't make any difference), although any \fBoverhead\fR
Packit d3f73b
value will be respected during calculations.
Packit d3f73b
.IP "atm, adsl"
Packit d3f73b
.br
Packit d3f73b
ATM linklayer consists of 53 byte cells, where each of them provides 48 bytes
Packit d3f73b
for payload. Also all the cells must be fully utilized, thus the last one is
Packit d3f73b
padded if/as necessary.
Packit d3f73b
Packit d3f73b
When the size table is calculated, adjusted size that fits properly into lowest
Packit d3f73b
amount of cells is assigned to a slot. For example, a 100 byte long packet
Packit d3f73b
requires three 48\-byte payloads, so the final size would require 3 ATM cells
Packit d3f73b
\- 159 bytes.
Packit d3f73b
Packit d3f73b
For ATM size tables, 16\~bytes sized slots are perfectly enough. The default
Packit d3f73b
values of \fBmtu\fR and \fBtsize\fR create 4\~bytes sized slots.
Packit d3f73b
.PP
Packit d3f73b
.
Packit d3f73b
.SH "TYPICAL OVERHEADS"
Packit d3f73b
The following values are typical for different adsl scenarios (based on
Packit d3f73b
\fB[1]\fR and \fB[2]\fR):
Packit d3f73b
Packit d3f73b
.nf
Packit d3f73b
LLC based:
Packit d3f73b
.RS 4
Packit d3f73b
PPPoA \- 14 (PPP \- 2, ATM \- 12)
Packit d3f73b
PPPoE \- 40+ (PPPoE \- 8, ATM \- 18, ethernet 14, possibly FCS \- 4+padding)
Packit d3f73b
Bridged \- 32 (ATM \- 18, ethernet 14, possibly FCS \- 4+padding)
Packit d3f73b
IPoA \- 16 (ATM \- 16)
Packit d3f73b
.RE
Packit d3f73b
Packit d3f73b
VC Mux based:
Packit d3f73b
.RS 4
Packit d3f73b
PPPoA \- 10 (PPP \- 2, ATM \- 8)
Packit d3f73b
PPPoE \- 32+ (PPPoE \- 8, ATM \- 10, ethernet 14, possibly FCS \- 4+padding)
Packit d3f73b
Bridged \- 24+ (ATM \- 10, ethernet 14, possibly FCS \- 4+padding)
Packit d3f73b
IPoA \- 8 (ATM \- 8)
Packit d3f73b
.RE
Packit d3f73b
.fi
Packit d3f73b
There are a few important things regarding the above overheads:
Packit d3f73b
.
Packit d3f73b
.IP \(bu 4
Packit d3f73b
IPoA in LLC case requires SNAP, instead of LLC\-NLPID (see rfc2684) \- this is
Packit d3f73b
the reason why it actually takes more space than PPPoA.
Packit d3f73b
.IP \(bu
Packit d3f73b
In rare cases, FCS might be preserved on protocols that include Ethernet frames
Packit d3f73b
(Bridged and PPPoE). In such situation, any Ethernet specific padding
Packit d3f73b
guaranteeing 64 bytes long frame size has to be included as well (see RFC2684).
Packit d3f73b
In the other words, it also guarantees that any packet you send will take
Packit d3f73b
minimum 2 atm cells. You should set \fBmpu\fR accordingly for that.
Packit d3f73b
.IP \(bu
Packit d3f73b
When the size table is consulted, and you're shaping traffic for the sake of
Packit d3f73b
another modem/router, an Ethernet header (without padding) will already be added
Packit d3f73b
to initial packet's length. You should compensate for that by subtracting 14
Packit d3f73b
from the above overheads in this case. If you're shaping directly on the router
Packit d3f73b
(for example, with speedtouch usb modem) using ppp daemon, you're using raw ip
Packit d3f73b
interface without underlying layer2, so nothing will be added.
Packit d3f73b
Packit d3f73b
For more thorough explanations, please see \fB[1]\fR and \fB[2]\fR.
Packit d3f73b
.
Packit d3f73b
.SH "ETHERNET CARDS CONSIDERATIONS"
Packit d3f73b
.
Packit d3f73b
It's often forgotten that modern network cards (even cheap ones on desktop
Packit d3f73b
motherboards) and/or their drivers often support different offloading
Packit d3f73b
mechanisms. In the context of traffic shaping, 'tso' and 'gso' might cause
Packit d3f73b
undesirable effects, due to massive TCP segments being considered during
Packit d3f73b
traffic shaping (including stab calculations). For slow uplink interfaces,
Packit d3f73b
it's good to use \fBethtool\fR to turn off offloading features.
Packit d3f73b
.
Packit d3f73b
.SH "SEE ALSO"
Packit d3f73b
.
Packit d3f73b
\fBtc\fR(8), \fBtc\-hfsc\fR(7), \fBtc\-hfsc\fR(8),
Packit d3f73b
.br
Packit d3f73b
\fB[1]\fR http://ace\-host.stuart.id.au/russell/files/tc/tc\-atm/
Packit d3f73b
.br
Packit d3f73b
\fB[2]\fR http://www.faqs.org/rfcs/rfc2684.html
Packit d3f73b
Packit d3f73b
Please direct bugreports and patches to: <netdev@vger.kernel.org>
Packit d3f73b
.
Packit d3f73b
.SH "AUTHOR"
Packit d3f73b
.
Packit d3f73b
Manpage created by Michal Soltys (soltys@ziu.info)