Penguin
Recent Changes
Annotated edit history of tc-prio(8) version 1, including all changes. View license author blame.
Rev Author # Line
1 perry 1 PRIO
2 !!!PRIO
3 NAME
4 SYNOPSIS
5 DESCRIPTION
6 ALGORITHM
7 CLASSIFICATION
8 QDISC PARAMETERS
9 CLASSES
10 BUGS
11 AUTHORS
12 ----
13 !!NAME
14
15
16 PRIO - Priority qdisc
17 !!SYNOPSIS
18
19
20 __tc qdisc ... dev__ dev __( parent__ classid __|
21 root) [[ handle__ major: __] prio [[ bands__ bands __] [[
22 priomap__ band,band,band... __] [[ estimator__ interval
23 timeconstant __]__
24 !!DESCRIPTION
25
26
27 The PRIO qdisc is a simple classful queueing discipline that
28 contains an arbitrary number of classes of differing
29 priority. The classes are dequeued in numerical descending
30 order of priority. PRIO is a scheduler and never delays
31 packets - it is a work-conserving qdisc, though the qdiscs
32 contained in the classes may not be.
33
34
35 Very useful for lowering latency when there is no need for
36 slowing down traffic.
37 !!ALGORITHM
38
39
40 On creation with 'tc qdisc add', a fixed number of bands is
41 created. Each band is a class, although is not possible to
42 add classes with 'tc qdisc add', the number of bands to be
43 created must instead be specified on the commandline
44 attaching PRIO to its root.
45
46
47 When dequeueing, band 0 is tried first and only if it did
48 not deliver a packet does PRIO try band 1, and so onwards.
49 Maximum reliability packets should therefore go to band 0,
50 minimum delay to band 1 and the rest to band 2.
51
52
53 As the PRIO qdisc itself will have minor number 0, band 0 is
54 actually major:1, band 1 is major:2, etc. For major,
55 substitute the major number assigned to the qdisc on 'tc
56 qdisc add' with the __handle__ parameter.
57 !!CLASSIFICATION
58
59
60 Three methods are available to PRIO to determine in which
61 band a packet will be enqueued.
62
63
64 From userspace
65
66
67 A process with sufficient privileges can encode the
68 destination class directly with SO_PRIORITY, see
69 __tc(7).__
70
71
72 with a tc filter
73
74
75 A tc filter attached to the root qdisc can point traffic
76 directly to a class
77
78
79 with the priomap
80
81
82 Based on the packet priority, which in turn is derived from
83 the Type of Service assigned to the packet.
84
85
86 Only the priomap is specific to this qdisc.
87 !!QDISC PARAMETERS
88
89
90 bands
91
92
93 Number of bands. If changed from the default of 3,
94 __priomap__ must be updated as well.
95
96
97 priomap
98
99
100 The priomap maps the priority of a packet to a class. The
101 priority can either be set directly from userspace, or be
102 derived from the Type of Service of the packet.
103
104
105 Determines how packet priorities, as assigned by the kernel,
106 map to bands. Mapping occurs based on the TOS octet of the
107 packet, which looks like this:
108
109
110 0 1 2 3 4 5 6 7
111 +---+---+---+---+---+---+---+---+
112 | | | |
113 |PRECEDENCE | TOS |MBZ|
114 | | | |
115 +---+---+---+---+---+---+---+---+
116 The four TOS bits (the 'TOS field') are defined as:
117
118
119 Binary Decimcal Meaning
120 -----------------------------------------
121 1000 8 Minimize delay (md)
122 0100 4 Maximize throughput (mt)
123 0010 2 Maximize reliability (mr)
124 0001 1 Minimize monetary cost (mmc)
125 0000 0 Normal Service
126 As there is 1 bit to the right of these four bits, the actual value of the TOS field is double the value of the TOS bits. Tcpdump -v -v shows you the value of the entire TOS field, not just the four bits. It is the value you see in the first column of this table:
127
128
129 TOS Bits Means Linux Priority Band
130 ------------------------------------------------------------
131 0x0 0 Normal Service 0 Best Effort 1
132 0x2 1 Minimize Monetary Cost 1 Filler 2
133 0x4 2 Maximize Reliability 0 Best Effort 1
134 0x6 3 mmc+mr 0 Best Effort 1
135 0x8 4 Maximize Throughput 2 Bulk 2
136 0xa 5 mmc+mt 2 Bulk 2
137 0xc 6 mr+mt 2 Bulk 2
138 0xe 7 mmc+mr+mt 2 Bulk 2
139 0x10 8 Minimize Delay 6 Interactive 0
140 0x12 9 mmc+md 6 Interactive 0
141 0x14 10 mr+md 6 Interactive 0
142 0x16 11 mmc+mr+md 6 Interactive 0
143 0x18 12 mt+md 4 Int. Bulk 1
144 0x1a 13 mmc+mt+md 4 Int. Bulk 1
145 0x1c 14 mr+mt+md 4 Int. Bulk 1
146 0x1e 15 mmc+mr+mt+md 4 Int. Bulk 1
147 The second column contains the value of the relevant four TOS bits, followed by their translated meaning. For example, 15 stands for a packet wanting Minimal Montetary Cost, Maximum Reliability, Maximum Throughput AND Minimum Delay.
148
149
150 The fourth column lists the way the Linux kernel interprets
151 the TOS bits, by showing to which Priority they are
152 mapped.
153
154
155 The last column shows the result of the default priomap. On
156 the commandline, the default priomap looks like
157 this:
158
159
160 1, 2, 2, 2, 1, 2, 0, 0 , 1, 1, 1, 1, 1, 1, 1, 1
161
162
163 This means that priority 4, for example, gets mapped to band
164 number 1. The priomap also allows you to list higher
165 priorities (
166
167
168 This table from RFC 1349 (read it for more details) explains
169 how applications might very well set their TOS
170 bits:
171
172
173 TELNET 1000 (minimize delay)
174 FTP
175 Control 1000 (minimize delay)
176 Data 0100 (maximize throughput)
177 TFTP 1000 (minimize delay)
178 SMTP
179 Command phase 1000 (minimize delay)
180 DATA phase 0100 (maximize throughput)
181 Domain Name Service
182 UDP Query 1000 (minimize delay)
183 TCP Query 0000
184 Zone Transfer 0100 (maximize throughput)
185 NNTP 0001 (minimize monetary cost)
186 ICMP
187 Errors 0000
188 Requests 0000 (mostly)
189 Responses
190 !!CLASSES
191
192
193 PRIO classes cannot be configured further - they are
194 automatically created when the PRIO qdisc is attached. Each
195 class however can contain yet a further qdisc.
196 !!BUGS
197
198
199 Large amounts of traffic in the lower bands can cause
200 starvation of higher bands. Can be prevented by attaching a
201 shaper (for example, __tc-tbf(8)__ to these bands to make
202 sure they cannot dominate the link.
203 !!AUTHORS
204
205
206 Alexey N. Kuznetsov,
207 ----
This page is a man page (or other imported legacy content). We are unable to automatically determine the license status of this page.