1NETEM(8) Linux NETEM(8)
2
6 NetEm - Network Emulator
7
9 tc qdisc ... dev DEVICE ] add netem OPTIONS
10 OPTIONS := [ LIMIT ] [ DELAY ] [ LOSS ] [ CORRUPT ] [ DUPLICATION ] [
12 REORDERING ] [ RATE ] [ SLOT ]
13 LIMIT := limit packets
15 DELAY := delay TIME [ JITTER [ CORRELATION ]]]
17 [ distribution { uniform | normal | pareto | paretonormal } ]
18 LOSS := loss { random PERCENT [ CORRELATION ] |
20 state p13 [ p31 [ p32 [ p23 [ p14]]]] |
21 gemodel p [ r [ 1-h [ 1-k ]]] } [ ecn ]
22 CORRUPT := corrupt PERCENT [ CORRELATION ]]
24 DUPLICATION := duplicate PERCENT [ CORRELATION ]]
26 REORDERING := reorder PERCENT [ CORRELATION ] [ gap DISTANCE ]
28 RATE := rate RATE [ PACKETOVERHEAD [ CELLSIZE [ CELLOVERHEAD ]]]]
30 SLOT := slot { MIN_DELAY [ MAX_DELAY ] |
32 distribution { uniform | normal | pareto | paretonormal
33 | FILE } DELAY JITTER }
34 [ packets PACKETS ] [ bytes BYTES ]
35
39 NetEm is an enhancement of the Linux traffic control facilities that
40 allow one to add delay, packet loss, duplication and more other charac‐
41 teristics to packets outgoing from a selected network interface. NetEm
42 is built using the existing Quality Of Service (QOS) and Differentiated
43 Services (diffserv) facilities in the Linux kernel.
44
47 netem has the following options:
48 limit packets
51 maximum number of packets the qdisc may hold queued at a time.
52 delay
55 adds the chosen delay to the packets outgoing to chosen network inter‐
56 face. The optional parameters allows one to introduce a delay variation
57 and a correlation. Delay and jitter values are expressed in ms while
58 correlation is percentage.
59 distribution
62 allow the user to choose the delay distribution. If not specified, the
63 default distribution is Normal. Additional parameters allow one to con‐
64 sider situations in which network has variable delays depending on
65 traffic flows concurring on the same path, that causes several delay
66 peaks and a tail.
67 loss random
70 adds an independent loss probability to the packets outgoing from the
71 chosen network interface. It is also possible to add a correlation, but
72 this option is now deprecated due to the noticed bad behavior.
73 loss state
76 adds packet losses according to the 4-state Markov using the transition
77 probabilities as input parameters. The parameter p13 is mandatory and
78 if used alone corresponds to the Bernoulli model. The optional parame‐
79 ters allows one to extend the model to 2-state (p31), 3-state (p23 and
80 p32) and 4-state (p14). State 1 corresponds to good reception, State 4
81 to independent losses, State 3 to burst losses and State 2 to good re‐
82 ception within a burst.
83 loss gemodel
86 adds packet losses according to the Gilbert-Elliot loss model or its
87 special cases (Gilbert, Simple Gilbert and Bernoulli). To use the
88 Bernoulli model, the only needed parameter is p while the others will
89 be set to the default values r=1-p, 1-h=1 and 1-k=0. The parameters
90 needed for the Simple Gilbert model are two (p and r), while three pa‐
91 rameters (p, r, 1-h) are needed for the Gilbert model and four (p, r,
92 1-h and 1-k) are needed for the Gilbert-Elliot model. As known, p and r
93 are the transition probabilities between the bad and the good states,
94 1-h is the loss probability in the bad state and 1-k is the loss proba‐
95 bility in the good state.
96 ecn
99 can be used optionally to mark packets instead of dropping them. A loss
100 model has to be used for this to be enabled.
101 corrupt
104 allows the emulation of random noise introducing an error in a random
105 position for a chosen percent of packets. It is also possible to add a
106 correlation through the proper parameter.
107 duplicate
110 using this option the chosen percent of packets is duplicated before
111 queuing them. It is also possible to add a correlation through the
112 proper parameter.
113 reorder
116 to use reordering, a delay option must be specified. There are two ways
117 to use this option (assuming 'delay 10ms' in the options list).
118 reorder 25% 50% gap 5
120 in this first example, the first 4 (gap - 1) packets are delayed by
121 10ms and subsequent packets are sent immediately with a probability of
122 0.25 (with correlation of 50% ) or delayed with a probability of 0.75.
123 After a packet is reordered, the process restarts i.e. the next 4 pack‐
124 ets are delayed and subsequent packets are sent immediately or delayed
125 based on reordering probability. To cause a repeatable pattern where
126 every 5th packet is reordered reliably, a reorder probability of 100%
127 can be used.
128 reorder 25% 50%
130 in this second example 25% of packets are sent immediately (with corre‐
131 lation of 50%) while the others are delayed by 10 ms.
132 rate
135 delay packets based on packet size and is a replacement for TBF. Rate
136 can be specified in common units (e.g. 100kbit). Optional PACKETOVER‐
137 HEAD (in bytes) specify an per packet overhead and can be negative. A
138 positive value can be used to simulate additional link layer headers. A
139 negative value can be used to artificial strip the Ethernet header
140 (e.g. -14) and/or simulate a link layer header compression scheme. The
141 third parameter - an unsigned value - specify the cellsize. Cellsize
142 can be used to simulate link layer schemes. ATM for example has an pay‐
143 load cellsize of 48 bytes and 5 byte per cell header. If a packet is 50
144 byte then ATM must use two cells: 2 * 48 bytes payload including 2 * 5
145 byte header, thus consume 106 byte on the wire. The last optional value
146 CELLOVERHEAD can be used to specify per cell overhead - for our ATM ex‐
147 ample 5. CELLOVERHEAD can be negative, but use negative values with
148 caution.
149 Note that rate throttling is limited by several factors: the kernel
151 clock granularity avoid a perfect shaping at a specific level. This
152 will show up in an artificial packet compression (bursts). Another in‐
153 fluence factor are network adapter buffers which can also add artifi‐
154 cial delay.
155 slot
158 defer delivering accumulated packets to within a slot. Each available
159 slot can be configured with a minimum delay to acquire, and an optional
160 maximum delay. Alternatively it can be configured with the distribu‐
161 tion similar to distribution for delay option. Slot delays can be spec‐
162 ified in nanoseconds, microseconds, milliseconds or seconds (e.g.
163 800us). Values for the optional parameters BYTES will limit the number
164 of bytes delivered per slot, and/or PACKETS will limit the number of
165 packets delivered per slot.
166 These slot options can provide a crude approximation of bursty MACs
168 such as DOCSIS, WiFi, and LTE.
169 Note that slotting is limited by several factors: the kernel clock
171 granularity, as with a rate, and attempts to deliver many packets
172 within a slot will be smeared by the timer resolution, and by the un‐
173 derlying native bandwidth also.
174 It is possible to combine slotting with a rate, in which case complex
176 behaviors where either the rate, or the slot limits on bytes or packets
177 per slot, govern the actual delivered rate.
178
181 The main known limitation of Netem are related to timer granularity,
182 since Linux is not a real-time operating system.
183
186 tc qdisc add dev eth0 root netem rate 5kbit 20 100 5
187 delay all outgoing packets on device eth0 with a rate of 5kbit, a
188 per packet overhead of 20 byte, a cellsize of 100 byte and a per
189 celloverhead of 5 byte:
190
193 1. Hemminger S. , "Network Emulation with NetEm", Open Source Develop‐
194 ment Lab, April 2005 (http://devresources.linux-founda‐
195 tion.org/shemminger/netem/LCA2005_paper.pdf)
196 2. Netem page from Linux foundation, (https://wiki.linuxfounda‐
199 tion.org/networking/netem)
200 3. Salsano S., Ludovici F., Ordine A., "Definition of a general and
203 intuitive loss model for packet networks and its implementation in
204 the Netem module in the Linux kernel", available at http://net‐
205 group.uniroma2.it/NetemCLG
206
209 tc(8), tc-tbf(8)
210
213 Netem was written by Stephen Hemminger at Linux foundation and is based
214 on NISTnet. This manpage was created by Fabio Ludovici <fabio.ludovici
215 at yahoo dot it> and Hagen Paul Pfeifer <hagen@jauu.net>
216iproute2 25 November 2011 NETEM(8)