3 <TITLE>Iperf version 2.0.0</TITLE>
4 <!-- $Id: index.html,v 1.1.1.1 2004/05/18 01:50:44 kgibbs Exp $ -->
7 <BODY BGCOLOR="#FFFFFF" LINK="#006633" VLINK="#669900" ALINK="#669966">
10 <P><IMG SRC="dast.gif"
11 ALT="Distributed Applications Support Team"></P>
14 <H1>Iperf version 2.0.0</H1>
18 <HR><!-- ----- ---- ---- ---- ---- ---- ---- ---- ---- ---- ---- ---- ---- -->
20 <H3>NLANR applications support
21 <BR><A HREF="http://dast.nlanr.net/">http://dast.nlanr.net/</A>
22 <BR><A HREF="mailto:dast@nlanr.net"><dast@nlanr.net></A>
24 <P><FONT face="arial,helvetica">
25 <H1>Iperf User Docs</H1>
33 [<a href="#compiling">Compiling</A> |
34 <A href="#features">Features</A> |
35 <A href="#tuningtcp">Tuning a TCP connection</A> |
36 <A href="#tuningudp">Tuning a UDP connection</A> |
37 <A href="#multicast">Running multicast servers and clients</A> |
38 <A href="#ipv6">IPv6 Mode</A> |
39 <A href="#repmode">Representative Streams</A> |
40 <A href="#daemon"> Running Iperf as a daemon</A> |
41 <!--<A href="#adaptive">Adaptive Window Sizes</A> | -->
42 <A href="#service">Running Iperf as a Windows Service</A> ]
44 <!-- ----- ---- ---- ---- ---- ---- ---- ---- ---- ---- ---- ---- ---- -->
45 <H2><A name=compiling></A>Compiling</H2>
47 Once you have the distribution, on UNIX,
48 unpack it using gzip and tar. That will create a new directory
49 'iperf-<version#>' with the source files and documentation.
52 Iperf compiles cleanly on many systems including Linux, SGI IRIX, HP-UX,
53 Solaris, AIX, and Cray UNICOS. Use '<TT>make</TT>' to configure for your OS and
54 compile the source code.
58 gunzip -c iperf-<version>.tar.gz | tar -xvf -
59 cd iperf-<version>
65 To install iperf, use '<TT>make install</TT>',
66 which will ask you where to install it. To recompile, the easiest way is to
67 start over. Do '<TT>make distclean</TT>' then '<TT>./configure; make</TT>'. See the Makefile
71 If you have problems, please report them to <A href="mailto:dast@nlanr.net">dast@nlanr.net</A> and
72 we will try to fix them quickly. <BR>
75 <!-- ----- ---- ---- ---- ---- ---- ---- ---- ---- ---- ---- ---- ---- -->
77 <H2><A name=features></A>Features</H2>
82 <LI>Report MSS/MTU size and observed read sizes.
83 <LI>Support for TCP window size via socket buffers.
84 <LI>Multi-threaded if pthreads or Win32 threads are available. Client and
85 server can have multiple simultaneous connections.
86 <!-- <LI>Suggest the optimal window size for a connection where the OS allows
87 setting window sizes in the granularity of bytes. </LI>--></UL>
90 <LI>Client can create UDP streams of specified bandwidth.
91 <LI>Measure packet loss
92 <LI>Measure delay jitter
94 <LI>Multi-threaded if pthreads are available. Client and server can have
95 multiple simultaneous connections. (This doesn't work in Windows.) </LI></UL>
96 <LI>Where appropriate, options can be specified with K (kilo-) and M (mega-)
97 suffices. So 128K instead of 131072 bytes.
98 <LI>Can run for specified time, rather than a set amount of data to transfer.
99 <LI>Picks the best units for the size of data being reported.
100 <LI>Server handles multiple connections, rather than quitting after a single
102 <LI>Print periodic, intermediate bandwidth, jitter, and loss reports at
104 <LI>Run the server as a daemon (Check out <A
105 href="http://www-itg.lbl.gov/nettest">Nettest</A> for running it as a secure
107 <LI>Run the server as a Windows NT Service
108 <LI>Use representative streams to test out how link layer compression affects
109 your achievable bandwidth.
111 <!-- <LI>A library of <A
112 href="lib.html">useful functions and C++
117 <!-- ----- ---- ---- ---- ---- ---- ---- ---- ---- ---- ---- ---- ---- --><BR>
119 <TABLE cellPadding=3 border=1>
122 <TH align=left>Command line option</TH>
123 <TH align=left>Environment variable option</TH>
124 <TH align=left>Description</TH></TR>
126 <TH bgColor=#cccccc colSpan=3>Client and Server options</TH></TR>
128 <TD><A name=format></A><TT>-f, --format <I>[bkmaBKMA]</I></TT></TD>
129 <TD><TT>$IPERF_FORMAT</TT></TD>
130 <TD>A letter specifying the format to print bandwidth numbers in.
131 Supported formats are
133 'b' = bits/sec 'B' = Bytes/sec
134 'k' = Kbits/sec 'K' = KBytes/sec
135 'm' = Mbits/sec 'M' = MBytes/sec
136 'g' = Gbits/sec 'G' = GBytes/sec
137 'a' = adaptive bits/sec 'A' = adaptive Bytes/sec
139 The adaptive formats choose between kilo- and mega- as appropriate. Fields
140 other than bandwidth always print bytes, but otherwise follow the
141 requested format. Default is 'a'. <BR><I>NOTE:</I> here Kilo = 1024,
142 Mega = 1024^2 and Giga = 1024^3 when dealing with bytes. Commonly in networking,
143 Kilo = 1000, Mega = 1000^2, and Giga = 1000^3 so we use this when dealing with
144 bits. If this really bothers you, use -f b and do the math.</TD></TR>
146 <TD><A name=interval></A><TT>-i, --interval <I>#</I></TT></TD>
147 <TD><TT>$IPERF_INTERVAL</TT></TD>
148 <TD>Sets the interval time in seconds between periodic bandwidth, jitter,
149 and loss reports. If non-zero, a report is made every <I>interval</I>
150 seconds of the bandwidth since the last report. If zero, no periodic
151 reports are printed. Default is zero.</TD></TR>
153 <TD><A name=len></A><TT>-l, --len <I>#[KM]</I></TT></TD>
154 <TD><TT>$IPERF_LEN</TT></TD>
155 <TD>The length of buffers to read or write. Iperf works by writing an
156 array of <I>len</I> bytes a number of times. Default is 8 KB for TCP, 1470
157 bytes for UDP. Note for UDP, this is the datagram size and needs to be lowered when using
158 IPv6 addressing to 1450 or less to avoid fragmentation. See also the <A
164 <TD><A name=print_mss></A><TT>-m, --print_mss</TT></TD>
165 <TD><TT>$IPERF_PRINT_MSS</TT></TD>
166 <TD>Print the reported TCP MSS size (via the TCP_MAXSEG option) and the
167 observed read sizes which often correlate with the MSS. The MSS is usually
168 the MTU - 40 bytes for the TCP/IP header. Often a slightly smaller MSS is
169 reported because of extra header space from IP options. The interface type
170 corresponding to the MTU is also printed (ethernet, FDDI, etc.). This
171 option is not implemented on many OSes, but the read sizes may still
172 indicate the MSS.</TD></TR>
174 <TD><A name=port></A><TT>-p, --port <I>#</I></TT></TD>
175 <TD><TT>$IPERF_PORT</TT></TD>
176 <TD>The server port for the server to listen on and the client to connect
177 to. This should be the same in both client and server. Default is 5001,
178 the same as ttcp.</TD></TR>
180 <TD><A name=udp></A><TT>-u, --udp</TT></TD>
181 <TD><TT>$IPERF_UDP</TT></TD>
182 <TD>Use UDP rather than TCP. See also the <A
183 href="#bandwidth">-b</A>
186 <TD><A name=window></A><TT>-w, --window <I>#[KM]</I></TT></TD>
187 <TD><TT>$TCP_WINDOW_SIZE</TT></TD>
188 <TD>Sets the socket buffer sizes to the specified value. For TCP, this
189 sets the TCP window size. For UDP it is just the buffer which datagrams
190 are received in, and so limits the largest receivable datagram size.</TD></TR>
192 <TD><A name=bind></A><TT>-B, --bind <I>host</I></TT></TD>
193 <TD><TT>$IPERF_BIND</TT></TD>
194 <TD>Bind to <I>host</I>, one of this machine's addresses. For the client
195 this sets the outbound interface. For a server this sets the incoming
196 interface. This is only useful on multihomed hosts, which have multiple
197 network interfaces.
198 <P>For Iperf in UDP server mode, this is also used to bind and join to a
199 multicast group. Use addresses in the range 224.0.0.0 to 239.255.255.255
200 for multicast. See also the <A
202 option.</P></TD></TR>
204 <TD><A name=compatibility></A><TT>-C, --compatibility </TT></TD>
205 <TD><TT>$IPERF_COMPAT</TT></TD>
206 <TD>Compatibility mode allows for use with older version of iperf. This mode
207 is not required for interoperability but it is highly recommended. In
208 some cases when using representative streaming you could cause a 1.7 server
209 to crash or cause undesired connection attempts.</P></TD></TR>
211 <TD><A name=mss></A><TT>-M, --mss <I>#[KM}</I></TT></TD>
212 <TD><TT>$IPERF_MSS</TT></TD>
213 <TD>Attempt to set the TCP maximum segment size (MSS) via the TCP_MAXSEG
214 option. The MSS is usually the MTU - 40 bytes for the TCP/IP header. For
215 ethernet, the MSS is 1460 bytes (1500 byte MTU). This option is not
216 implemented on many OSes.</TD></TR>
218 <TD><A name=nodelay></A><TT>-N, --nodelay</TT></TD>
219 <TD><TT>$IPERF_NODELAY</TT></TD>
220 <TD>Set the TCP no delay option, disabling Nagle's algorithm. Normally
221 this is only disabled for interactive applications like telnet.</TD></TR>
223 <TD><TT>-V </TT>(from v1.6 or higher)</TD>
225 <TD>Bind to an IPv6 address <BR>Server side: <BR>$ iperf -s -V
226 <P>Client side: <BR>$ iperf -c <Server IPv6 Address> -V
227 <BR> </P>Note: On version 1.6.3 and later a specific IPv6 Address does
228 not need to be bound with the <A href="#bind">-B</A> option, previous 1.6
229 versions do. Also on most OSes using this option will also respond to IPv4
230 clients using IPv4 mapped addresses.</TD></TR>
232 <TH bgColor=#cccccc colSpan=3>Server specific options</TH></TR>
234 <TD><A name=server></A><TT>-s, --server</TT></TD>
235 <TD><TT>$IPERF_SERVER</TT></TD>
236 <TD>Run Iperf in server mode.</TD></TR>
238 <TD><TT>-D </TT> (from v1.2 or higher)</TD>
240 <TD>Run the server as a daemon (Unix platforms) <BR>On Win32 platforms
241 where services are available, Iperf will start running as a service.</TD></TR>
243 <TD><TT>-R </TT>(only for Windows, from v1.2 or higher)</TD>
245 <TD>Remove the Iperf service (if it's running). </TD></TR><TR>
246 <TD><TT>-o </TT>(only for Windows, from v1.2 or higher)</TD>
248 <TD>Redirect output to given file. </TD></TR>
250 <TD><A name=sclient></A><TT>-c, --client <I>host</I></TT></TD>
251 <TD><TT>$IPERF_CLIENT</TT></TD>
252 <TD> If Iperf is in server mode, then specifying a host with -c
253 will limit the connections that Iperf will accept to the
254 <I>host</I> specified. Does not work well for UDP.</TD></TR>
256 <TD><A name=sparallel></A><TT>-P, --parallel <I>#</I></TT></TD>
257 <TD><TT>$IPERF_PARALLEL</TT></TD>
258 <TD>The number of connections to handle by the server before
259 closing. Default is 0 (which means to accept connections forever).</TD></TR>
261 <TH bgColor=#cccccc colSpan=3>Client specific options</TH></TR>
263 <TD><A name=bandwidth></A><TT>-b, --bandwidth <I>#[KM]</I></TT></TD>
264 <TD><TT>$IPERF_BANDWIDTH</TT></TD>
265 <TD>The UDP bandwidth to send at, in bits/sec. This implies the -u option.
266 Default is 1 Mbit/sec.</TD></TR>
268 <TD><A name=client></A><TT>-c, --client <I>host</I></TT></TD>
269 <TD><TT>$IPERF_CLIENT</TT></TD>
270 <TD>Run Iperf in client mode, connecting to an Iperf server running on
271 <I>host</I>.</TD></TR>
273 <TD><A name=dualtest></A><TT>-d, --dualtest </TT></TD>
274 <TD><TT>$IPERF_DUALTEST</TT></TD>
275 <TD>Run Iperf in dual testing mode. This will cause the server to connect
276 back to the client on the port specified in the
277 <A href="#listenport">-L</A> option (or defaults
278 to the port the client connected to the server on). This is done immediately
279 therefore running the tests simultaneously. If you want an alternating
280 test try <A href="#tradeoff">-r.</A></TD></TR>
282 <TD><A name=num></A><TT>-n, --num <I>#[KM]</I></TT></TD>
283 <TD><TT>$IPERF_NUM</TT></TD>
284 <TD>The number of buffers to transmit. Normally, Iperf sends for 10
285 seconds. The -n option overrides this and sends an array of <I>len</I>
286 bytes <I>num</I> times, no matter how long that takes. See also the <A
292 <TD><A name=tradeoff></A><TT>-r, --tradeoff </TT></TD>
293 <TD><TT>$IPERF_TRADEOFF</TT></TD>
294 <TD>Run Iperf in tradeoff testing mode. This will cause the server to connect
295 back to the client on the port specified in the
296 <A href="#listenport">-L</A> option (or defaults
297 to the port the client connected to the server on). This is done following
298 the client connection termination, therefore running the tests
299 alternating. If you want an simultaneous test try
300 <A href="#dualtest">-d.</A></TD></TR>
302 <TD><A name=time></A><TT>-t, --time <I>#</I></TT></TD>
303 <TD><TT>$IPERF_TIME</TT></TD>
304 <TD>The time in seconds to transmit for. Iperf normally works by
305 repeatedly sending an array of <I>len</I> bytes for <I>time</I> seconds.
306 Default is 10 seconds. See also the <A
312 <TD><A name=listenport></A><TT>-L, --listenport <I>#</I></TT></TD>
313 <TD><TT>$IPERF_LISTENPORT</TT></TD>
314 <TD>This specifies the port that the server will connect back to the
315 client on. It defaults to the port used to connect to the server
316 from the client.</TD></TR>
318 <TD><A name=parallel></A><TT>-P, --parallel <I>#</I></TT></TD>
319 <TD><TT>$IPERF_PARALLEL</TT></TD>
320 <TD>The number of simultaneous connections to make to the server. Default
321 is 1. Requires thread support on both the client and server.</TD></TR>
323 <TD><A name=tos></A><TT>-S, --tos <I>#</I></TT></TD>
324 <TD><TT>$IPERF_TOS</TT></TD>
325 <TD>The type-of-service for outgoing packets. (Many routers ignore the TOS
326 field.) You may specify the value in hex with a '0x' prefix, in octal with
327 a '0' prefix, or in decimal. For example, '0x10' hex = '020' octal = '16'
328 decimal. The TOS numbers specified in RFC 1349 are:
330 IPTOS_LOWDELAY minimize delay 0x10
331 IPTOS_THROUGHPUT maximize throughput 0x08
332 IPTOS_RELIABILITY maximize reliability 0x04
333 IPTOS_LOWCOST minimize cost 0x02
334
338 <TD><A name=ttl></A><TT>-T, --ttl <I>#</I></TT></TD>
339 <TD><TT>$IPERF_TTL</TT></TD>
340 <TD>The time-to-live for outgoing multicast packets. This is essentially
341 the number of router hops to go through, and is also used for scoping.
342 Default is 1, link-local.</TD></TR>
344 <TD><TT>-F</TT> (from v1.2 or higher)</TD>
346 <TD>Use a representative stream to measure bandwidth, e.g. :- <BR>$
347 iperf -c <server address> -F <file-name></TD></TR>
349 <TD><TT>-I </TT>(from v1.2 or higher)</TD>
351 <TD>Same as -F, input from stdin.</TD></TR>
353 <TD><TT>-W </TT>(from v1.2 or higher)</TD>
355 <TD>Adaptive Window Sizes.
356 <BR>Use Iperf to suggest the best Window size for a connection. Iperf will start from a default window size and try to perform a search for the optimal window size</TD></TR>
358 <TH bgColor=#cccccc colSpan=3>Miscellaneous options</TH></TR>
360 <TD><A name=help></A><TT>-h, --help</TT></TD>
362 <TD>Print out a summary of commands and quit.</TD></TR>
364 <TD><A name=version></A><TT>-v, --version</TT></TD>
366 <TD>Print version information and quit. Prints 'pthreads' if compiled with
367 POSIX threads, 'win32 threads' if compiled with Microsoft Win32 threads,
368 or 'single threaded' if compiled without threads.</TD></TR></TBODY></TABLE>
371 <!-- ----- ---- ---- ---- ---- ---- ---- ---- ---- ---- ---- ---- ---- -->
372 <H2><A name=tuningtcp></A>Tuning a TCP connection</H2>
374 The primary goal of Iperf
375 is to help in tuning TCP connections over a particular path. The most
376 fundamental tuning issue for TCP is the TCP window size, which controls how much
377 data can be in the network at any one point. If it is too small, the sender will
378 be idle at times and get poor performance. The theoretical value to use for the
379 TCP window size is the <I>bandwidth delay product</I>,
380 <BLOCKQUOTE>bottleneck bandwidth * round trip time</BLOCKQUOTE>In the below
381 modi4/cyclops example, the bottleneck link is a 45 Mbit/sec DS3 link and the
382 round trip time measured with ping is 42 ms. The bandwidth delay product is
383 <BLOCKQUOTE>45 Mbit/sec * 42 ms <BR>= (45e6) * (42e-3) <BR>= 1890000 bits
384 <BR>= 230 KByte</BLOCKQUOTE>That is a starting point for figuring the best
385 window size; setting it higher or lower may produce better results. In our
386 example, buffer sizes over 130K did not improve the performance, despite the
387 bandwidth delay product of 230K.
388 <P>Note that many OSes and hosts have upper limits on the TCP window size. These
389 may be as low as 64 KB, or as high as several MB. Iperf tries to detect when
390 these occur and give a warning that the actual and requested window sizes are
391 not equal (as below, though that is due to rounding in IRIX). PSC has a <A
392 href="http://www.psc.edu/networking/perf_tune.html">list detailing</A> how to
393 change the default and maximum window sizes for various OSes. For more
394 information on TCP window sizes, see the <A
395 href="http://dast.nlanr.net/Guides/GettingStarted/TCP_window_size.html">User's
396 Guide to TCP Windows.</A>
397 <P>Here is an example session, between node1 in Illinois and node2 in North
398 Carolina. These are connected via the vBNS backbone and a 45 Mbit/sec DS3 link.
399 Notice we improve bandwidth performance by a factor of 3 using proper TCP window
400 sizes. Use the adaptive window sizes feature on platforms which allow setting
401 window sizes in the granularity of bytes.
404 <B>node2></B> iperf -s
405 ------------------------------------------------------------
406 Server listening on TCP port 5001
407 TCP window size: 60.0 KByte (default)
408 ------------------------------------------------------------
409 [ 4] local <IP Addr node2> port 5001 connected with <IP Addr node1> port 2357
410 [ ID] Interval Transfer Bandwidth
411 [ 4] 0.0-10.1 sec 6.5 MBytes <B><FONT color=#ff0000>5.2 Mbits/sec
413 </FONT>node1></B> iperf -c node2
414 ------------------------------------------------------------
415 Client connecting to node1, TCP port 5001
416 TCP window size: 59.9 KByte (default)
417 ------------------------------------------------------------
418 [ 3] local <IP Addr node1> port 2357 connected with <IP Addr node2> port 5001
419 [ ID] Interval Transfer Bandwidth
420 [ 3] 0.0-10.0 sec 6.5 MBytes 5.2 Mbits/sec</PRE>
422 <PRE><B>node2></B> iperf -s -w 130k
423 ------------------------------------------------------------
424 Server listening on TCP port 5001
425 TCP window size: 130 KByte
426 ------------------------------------------------------------
427 [ 4] local <IP Addr node 2> port 5001 connected with <IP Addr node 1> port 2530
428 [ ID] Interval Transfer Bandwidth
429 [ 4] 0.0-10.1 sec 19.7 MBytes <B><FONT color=#ff0000>15.7 Mbits/sec
431 </FONT>node1></B> iperf -c node2 -w 130k
432 ------------------------------------------------------------
433 Client connecting to node2, TCP port 5001
434 TCP window size: 129 KByte (WARNING: requested 130 KByte)
435 ------------------------------------------------------------
436 [ 3] local <IP Addr node1> port 2530 connected with <IP Addr node2> port 5001
437 [ ID] Interval Transfer Bandwidth
438 [ 3] 0.0-10.0 sec 19.7 MBytes 15.8 Mbits/sec</PRE></BLOCKQUOTE>Another
439 test to do is run parallel TCP streams. If the total aggregate bandwidth is more
440 than what an individual stream gets, something is wrong. Either the TCP window
441 size is too small, or the OS's TCP implementation has bugs, or the network
442 itself has deficiencies. See above for TCP window sizes; otherwise diagnosing
443 which is somewhat difficult. If Iperf is compiled with pthreads, a single client
444 and server can test this, otherwise setup multiple clients and servers on
445 different ports. Here's an example where a single stream gets 16.5 Mbit/sec, but
446 two parallel streams together get 16.7 + 9.4 = 26.1 Mbit/sec, even when using
447 large TCP window sizes:
448 <BLOCKQUOTE><PRE><B>node2></B> iperf -s -w 300k
449 ------------------------------------------------------------
450 Server listening on TCP port 5001
451 TCP window size: 300 KByte
452 ------------------------------------------------------------
453 [ 4] local <IP Addr node2> port 5001 connected with <IP Addr node1> port 6902
454 [ ID] Interval Transfer Bandwidth
455 [ 4] 0.0-10.2 sec 20.9 MBytes <B><FONT color=#ff0000>16.5 Mbits/sec
457 </FONT></B>[ 4] local <IP Addr node2> port 5001 connected with <IP Addr node1> port 6911
458 [ 5] local <IP Addr node2> port 5001 connected with <IP Addr node2> port 6912
459 [ ID] Interval Transfer Bandwidth
460 [ 5] 0.0-10.1 sec 21.0 MBytes <B><FONT color=#ff0000>16.7 Mbits/sec
461 </FONT></B>[ 4] 0.0-10.3 sec 12.0 MBytes <B><FONT color=#ff0000> 9.4 Mbits/sec
463 </FONT>node1></B> ./iperf -c node2 -w 300k
464 ------------------------------------------------------------
465 Client connecting to node2, TCP port 5001
466 TCP window size: 299 KByte (WARNING: requested 300 KByte)
467 ------------------------------------------------------------
468 [ 3] local <IP Addr node2> port 6902 connected with <IP Addr node1> port 5001
469 [ ID] Interval Transfer Bandwidth
470 [ 3] 0.0-10.2 sec 20.9 MBytes 16.4 Mbits/sec
472 <B>node1></B> iperf -c node2 -w 300k -P 2
473 ------------------------------------------------------------
474 Client connecting to node2, TCP port 5001
475 TCP window size: 299 KByte (WARNING: requested 300 KByte)
476 ------------------------------------------------------------
477 [ 4] local <IP Addr node2> port 6912 connected with <IP Addr node1> port 5001
478 [ 3] local <IP Addr node2> port 6911 connected with <IP Addr node1> port 5001
479 [ ID] Interval Transfer Bandwidth
480 [ 4] 0.0-10.1 sec 21.0 MBytes 16.6 Mbits/sec
481 [ 3] 0.0-10.2 sec 12.0 MBytes 9.4 Mbits/sec</PRE></BLOCKQUOTE>A
482 secondary tuning issue for TCP is the maximum transmission unit (MTU). To be
483 most effective, both hosts should support Path MTU Discovery. PSC has a <A
484 href="http://www.psc.edu/networking/perf_tune.html">list detailing</A> what OSes
485 support Path MTU Discovery. Hosts without Path MTU Discovery often use 536 as
486 the MSS, which wastes bandwidth and processing time. Use the -m option to
487 display what MSS is being used, and see if this matches what you expect. Often
488 it is around 1460 bytes for ethernet.
489 <BLOCKQUOTE><PRE><B>node3></B> iperf -s -m
490 ------------------------------------------------------------
491 Server listening on TCP port 5001
492 TCP window size: 60.0 KByte (default)
493 ------------------------------------------------------------
494 [ 4] local <IP Addr node3> port 5001 connected with <IP Addr node4> port 1096
495 [ ID] Interval Transfer Bandwidth
496 [ 4] 0.0- 2.0 sec 1.8 MBytes 6.9 Mbits/sec
497 [ 4] <B><FONT color=#ff0000>MSS size 1448 bytes (MTU 1500 bytes, ethernet)
498 </FONT></B>[ 4] Read lengths occurring in more than 5% of reads:
499 [ 4] 952 bytes read 219 times (16.2%)
500 [ 4] 1448 bytes read 1128 times (83.6%)</PRE></BLOCKQUOTE>Here
501 is a host that doesn't support Path MTU Discovery. It will only send and receive
502 small 576 byte packets.
503 <BLOCKQUOTE><PRE><B>node4></B> iperf -s -m
504 ------------------------------------------------------------
505 Server listening on TCP port 5001
506 TCP window size: 32.0 KByte (default)
507 ------------------------------------------------------------
508 [ 4] local <IP Addr node4> port 5001 connected with <IP Addr node3> port 13914
509 [ ID] Interval Transfer Bandwidth
510 [ 4] 0.0- 2.3 sec 632 KBytes 2.1 Mbits/sec
511 <B><FONT color=#ff0000>WARNING: Path MTU Discovery may not be enabled.
512 </FONT></B>[ 4] <B><FONT color=#ff0000>MSS size 536 bytes (MTU 576 bytes, minimum)
513 </FONT></B>[ 4] Read lengths occurring in more than 5% of reads:
514 [ 4] 536 bytes read 308 times (58.4%)
515 [ 4] 1072 bytes read 91 times (17.3%)
516 [ 4] 1608 bytes read 29 times (5.5%)</PRE></BLOCKQUOTE>Iperf
517 supports other tuning options, which were added for exceptional network
518 situations like HIPPI-to-HIPPI over ATM. <BR>
520 <!-- ----- ---- ---- ---- ---- ---- ---- ---- ---- ---- ---- ---- ---- -->
522 <H2><A name=tuningudp></A>Tuning a UDP connection</H2>
524 Iperf creates a constant bit rate UDP stream. This is a very artificial
525 stream, similar to voice communication but not much else.
528 You will want to adjust the datagram size (-l) to the size your application
532 The server detects UDP datagram loss by ID numbers in the datagrams. Usually
533 a UDP datagram becomes several IP packets. Losing a single IP packet will lose
534 the entire datagram. To measure packet loss instead of datagram loss, make the
535 datagrams small enough to fit into a single packet, using the -l option. The
536 default size of 1470 bytes works for ethernet. Out-of-order packets are also
537 detected. (Out-of-order packets cause some ambiguity in the lost packet count;
538 Iperf assumes they are not duplicate packets, so they are excluded from the lost
539 packet count.) Since TCP does not report loss to the user, I find UDP tests
540 helpful to see packet loss along a path.
543 Jitter calculations are continuously computed by the server, as specified by
544 RTP in RFC 1889. The client records a 64 bit second/microsecond timestamp in the
545 packet. The server computes the relative transit time as (server's receive time
546 - client's send time). The client's and server's clocks do not need to be
547 synchronized; any difference is subtracted out in the jitter calculation. Jitter
548 is the smoothed mean of differences between consecutive transit times.
549 <BLOCKQUOTE><PRE><B>node2></B> iperf -s -u -i 1
550 ------------------------------------------------------------
551 Server listening on UDP port 5001
552 Receiving 1470 byte datagrams
553 UDP buffer size: 60.0 KByte (default)
554 ------------------------------------------------------------
555 [ 4] local <IP Addr node2> port 5001 connected with <IP Addr node1> port 9726
556 [ ID] Interval Transfer Bandwidth Jitter Lost/Total Datagrams
557 [ 4] 0.0- 1.0 sec 1.3 MBytes 10.0 Mbits/sec 0.209 ms 1/ 894 (0.11%)
558 [ 4] 1.0- 2.0 sec 1.3 MBytes 10.0 Mbits/sec 0.221 ms 0/ 892 (0%)
559 [ 4] 2.0- 3.0 sec 1.3 MBytes 10.0 Mbits/sec 0.277 ms 0/ 892 (0%)
560 [ 4] 3.0- 4.0 sec 1.3 MBytes 10.0 Mbits/sec 0.359 ms 0/ 893 (0%)
561 [ 4] 4.0- 5.0 sec 1.3 MBytes 10.0 Mbits/sec 0.251 ms 0/ 892 (0%)
562 [ 4] 5.0- 6.0 sec 1.3 MBytes 10.0 Mbits/sec 0.215 ms 0/ 892 (0%)
563 [ 4] 6.0- 7.0 sec 1.3 MBytes 10.0 Mbits/sec 0.325 ms 0/ 892 (0%)
564 [ 4] 7.0- 8.0 sec 1.3 MBytes 10.0 Mbits/sec 0.254 ms 0/ 892 (0%)
565 [ 4] 8.0- 9.0 sec 1.3 MBytes 10.0 Mbits/sec 0.282 ms 0/ 892 (0%)
566 [ 4] 0.0-10.0 sec 12.5 MBytes 10.0 Mbits/sec 0.243 ms 1/ 8922 (0.011%)
568 <B>node1></B> iperf -c node2 -u -b 10m
569 ------------------------------------------------------------
570 Client connecting to node2, UDP port 5001
571 Sending 1470 byte datagrams
572 UDP buffer size: 60.0 KByte (default)
573 ------------------------------------------------------------
574 [ 3] local <IP Addr node1> port 9726 connected with <IP Addr node2> port 5001
575 [ ID] Interval Transfer Bandwidth
576 [ 3] 0.0-10.0 sec 12.5 MBytes 10.0 Mbits/sec
577 [ 3] Sent 8922 datagrams</PRE></BLOCKQUOTE>Notice the higher jitter due to
578 datagram reassembly when using larger 32 KB datagrams, each split into 23
579 packets of 1500 bytes. The higher datagram loss seen here may be due to the
580 burstiness of the traffic, which is 23 back-to-back packets and then a long
581 pause, rather than evenly spaced individual packets.
582 <BLOCKQUOTE><PRE><B>node2></B> iperf -s -u -l 32k -w 128k -i 1
583 ------------------------------------------------------------
584 Server listening on UDP port 5001
585 Receiving 32768 byte datagrams
586 UDP buffer size: 128 KByte
587 ------------------------------------------------------------
588 [ 3] local <IP Addr node2> port 5001 connected with <IP Addr node1> port 11303
589 [ ID] Interval Transfer Bandwidth Jitter Lost/Total Datagrams
590 [ 3] 0.0- 1.0 sec 1.3 MBytes 10.0 Mbits/sec 0.430 ms 0/ 41 (0%)
591 [ 3] 1.0- 2.0 sec 1.1 MBytes 8.5 Mbits/sec 5.996 ms 6/ 40 (15%)
592 [ 3] 2.0- 3.0 sec 1.2 MBytes 9.7 Mbits/sec 0.796 ms 1/ 40 (2.5%)
593 [ 3] 3.0- 4.0 sec 1.2 MBytes 10.0 Mbits/sec 0.403 ms 0/ 40 (0%)
594 [ 3] 4.0- 5.0 sec 1.2 MBytes 10.0 Mbits/sec 0.448 ms 0/ 40 (0%)
595 [ 3] 5.0- 6.0 sec 1.2 MBytes 10.0 Mbits/sec 0.464 ms 0/ 40 (0%)
596 [ 3] 6.0- 7.0 sec 1.2 MBytes 10.0 Mbits/sec 0.442 ms 0/ 40 (0%)
597 [ 3] 7.0- 8.0 sec 1.2 MBytes 10.0 Mbits/sec 0.342 ms 0/ 40 (0%)
598 [ 3] 8.0- 9.0 sec 1.2 MBytes 10.0 Mbits/sec 0.431 ms 0/ 40 (0%)
599 [ 3] 9.0-10.0 sec 1.2 MBytes 10.0 Mbits/sec 0.407 ms 0/ 40 (0%)
600 [ 3] 0.0-10.0 sec 12.3 MBytes 9.8 Mbits/sec 0.407 ms 7/ 401 (1.7%)
602 <B>node1></B> iperf -c node2 -b 10m -l 32k -w 128k
603 ------------------------------------------------------------
604 Client connecting to node2, UDP port 5001
605 Sending 32768 byte datagrams
606 UDP buffer size: 128 KByte
607 ------------------------------------------------------------
608 [ 3] local <IP Addr node2> port 11303 connected with <IP Addr node1> port 5001
609 [ ID] Interval Transfer Bandwidth
610 [ 3] 0.0-10.0 sec 12.5 MBytes 10.0 Mbits/sec
611 [ 3] Sent 401 datagrams</PRE><PRE></PRE></BLOCKQUOTE>
612 <P><A name=multicast></A>
613 <B><FONT size=+1>Multicast</FONT></B>
614 <P>To test multicast, run several servers with the bind option (-B, --bind) set
615 to the multicast group address. Run the client, connecting to the multicast
616 group address and setting the TTL (-T, --ttl) as needed. Unlike normal TCP and
617 UDP tests, multicast servers may be started after the client. In that case,
618 datagrams sent before the server started show up as losses in the first periodic
619 report (61 datagrams on arno below).
620 <BLOCKQUOTE><PRE><B>node5></B> iperf -c 224.0.67.67 -u --ttl 5 -t 5
621 ------------------------------------------------------------
622 Client connecting to 224.0.67.67, UDP port 5001
623 Sending 1470 byte datagrams
624 Setting multicast TTL to 5
625 UDP buffer size: 32.0 KByte (default)
626 ------------------------------------------------------------
627 [ 3] local <IP Addr node5> port 1025 connected with 224.0.67.67 port 5001
628 [ ID] Interval Transfer Bandwidth
629 [ 3] 0.0- 5.0 sec 642 KBytes 1.0 Mbits/sec
630 [ 3] Sent 447 datagrams
632 <B>node5></B> iperf -s -u -B 224.0.67.67 -i 1
633 ------------------------------------------------------------
634 Server listening on UDP port 5001
635 Binding to local address 224.0.67.67
636 Joining multicast group 224.0.67.67
637 Receiving 1470 byte datagrams
638 UDP buffer size: 32.0 KByte (default)
639 ------------------------------------------------------------
640 [ 3] local 224.0.67.67 port 5001 connected with <IP Addr node5> port 1025
641 [ ID] Interval Transfer Bandwidth Jitter Lost/Total Datagrams
642 [ 3] 0.0- 1.0 sec 131 KBytes 1.0 Mbits/sec 0.007 ms 0/ 91 (0%)
643 [ 3] 1.0- 2.0 sec 128 KBytes 1.0 Mbits/sec 0.008 ms 0/ 89 (0%)
644 [ 3] 2.0- 3.0 sec 128 KBytes 1.0 Mbits/sec 0.010 ms 0/ 89 (0%)
645 [ 3] 3.0- 4.0 sec 128 KBytes 1.0 Mbits/sec 0.013 ms 0/ 89 (0%)
646 [ 3] 4.0- 5.0 sec 128 KBytes 1.0 Mbits/sec 0.008 ms 0/ 89 (0%)
647 [ 3] 0.0- 5.0 sec 642 KBytes 1.0 Mbits/sec 0.008 ms 0/ 447 (0%)
649 <B>node6></B> iperf -s -u -B 224.0.67.67 -i 1
650 ------------------------------------------------------------
651 Server listening on UDP port 5001
652 Binding to local address 224.0.67.67
653 Joining multicast group 224.0.67.67
654 Receiving 1470 byte datagrams
655 UDP buffer size: 60.0 KByte (default)
656 ------------------------------------------------------------
657 [ 3] local 224.0.67.67 port 5001 connected with <IP Addr node5> port 1025
658 [ ID] Interval Transfer Bandwidth Jitter Lost/Total Datagrams
659 [ 3] 0.0- 1.0 sec 129 KBytes 1.0 Mbits/sec 0.778 ms 61/ 151 (40%)
660 [ 3] 1.0- 2.0 sec 128 KBytes 1.0 Mbits/sec 0.236 ms 0/ 89 (0%)
661 [ 3] 2.0- 3.0 sec 128 KBytes 1.0 Mbits/sec 0.264 ms 0/ 89 (0%)
662 [ 3] 3.0- 4.0 sec 128 KBytes 1.0 Mbits/sec 0.248 ms 0/ 89 (0%)
663 [ 3] 0.0- 4.3 sec 554 KBytes 1.0 Mbits/sec 0.298 ms 61/ 447 (14%)</PRE><PRE><HR width="100%"></PRE></BLOCKQUOTE>
666 <DT><B><FONT size=+2>IPv6 Mode</FONT></B>
667 <DD>Download the IPv6 version of this release.<BR>Get the IPv6 address of the node using the 'ifconfig' command.<BR>Use the <FONT color=#000099>-V</FONT> option to indicate that you are using an IPv6 address Please note that we need to explicitly bind the server address also.
668 <P>Server side:<BR><FONT color=#000099> $ iperf -s -V</FONT>
669 <P>Client side:<BR><FONT color=#000099>$ iperf -c <Server IPv6 Address> -V </FONT>
670 <P>Note: Iperf version 1.6.2 and eariler require a IPv6 address to be explicitly bound
671 with the <A HREF="#bind">-B</A> option for the server.</P></DD></DL>
673 <P><A name=repmode></A>
675 <DT><B><FONT size=+2>Using Representative Streams to measure
677 <DD>Use the -F or -I option. If you want to test how your network performs
678 with compressed / uncompressed streams, just create representative streams and
679 use the -F option to test it. This is usually due to the link layer
681 <P>The -F option is for file input.<BR>The -I option is for input from stdin.
682 <P>E.g. <BR>Client: $ <FONT color=#000099> iperf -c <server address> -F <file-name><BR></FONT>
683 <BR>Client: $ <FONT color=#000099> iperf -c <server address> -I </FONT></P></DD></DL>
684 <P><A name=daemon></A>
687 <DT><B><FONT size=+2>Running the server as a daemon</FONT></B>
688 <DD>Use the -D command line option to run the server as a daemon. Redirect the
689 output to a file.<BR>E.g. <FONT color=#000099>iperf -s -D >
690 iperfLog</FONT>. <FONT color=#000000>This will have the Iperf Server running
691 as a daemon and the server messages will be logged in the file iperfLog.
694 <P><A name=service></A>
696 <DT><B><FONT size=+2>Using Iperf as a Service under Win32</FONT></B>
697 <DD>There are three options for Win32:
700 <DT>-o outputfilename
701 <DD>output the messages into the specified file
703 <DD>install Iperf as a service and run it
705 <DD>uninstall the Iperf service </DD></DL>
708 <DT><FONT color=#3366ff>iperf -s -D -o iperflog.txt</FONT>
709 <DD>will install the Iperf service and run it. Messages will be reported
710 into "%windir%\system32\iperflog.txt"
712 <DT><FONT color=#3366ff>iperf -s -R</FONT>
713 <DD>will uninstall the Iperf service if it is installed.
714 <P>Note: If you stop want to restart the Iperf service after having killed
715 it with the Microsoft Management Console or the Windows Task Manager, make
716 sure to use the proper OPTION in the service properties dialog.
717 </P></DD></DL></DD></DL>
719 <!--<P><A name=multicast></A>
721 <DT><B><FONT size=+2>Running the multicast server and client</FONT></B>
722 <DD>Use the -B option while starting the server to bind it to a multicast
723 address.<BR>E.g. :-<FONT color=#3366ff>iperf -s -u -B 224.0.55.55</FONT>.
725 <P>This will have the Iperf server listening for datagrams (-u) for the
726 address 224.0.55.55(-B 224.0.55.55).
727 <P>Now, start a client sending packets to this multicast address.
728 <P>E.g. : <FONT color=#3366ff>iperf -c 224.0.55.55 -u</FONT>.
729 This will have a UDP client (-u) sending to the multicast address
730 224.0.55.55(-c 224.0.55.55).
731 <P><FONT color=#000000>Start multiple clients or servers as explained above,
732 sending data to the same multicast server. (If you have multiple servers
733 listening on the multicast address, each of the servers will be getting the data)
736 <!--<A name=adaptive></A>
738 <DT><B><FONT size=+2>Adaptive window sizes</FONT></B>
739 <DD>Use the -W option on the client to run the client with the adaptive window
740 size. Ensure that the server window size is fairly big for this
741 option.<BR>E.g.. If the server TCP window size is 8KB, it does not help having
742 a client TCP window size of 256KB.<BR>256KB Server TCP Window Size should
743 suffice for most high bandwidth networks.
744 <P>Client changes the TCP window size using a binary exponential
745 algorithm. This means that you may notice that TCP window size suggested may
746 vary according to the traffic in the network, Iperf will suggest the best
747 window size for the current network scenario.
750 --><P><!-- ----- ---- ---- ---- ---- ---- ---- ---- ---- ---- ---- ---- ---- -->
752 <P>Copyright 1999,2000,2001,2002,2003,2004 <BR>The Board of Trustees of the University of
753 Illinois <BR>All rights reserved <BR>See <A
754 href="ui_license.html">UI License</A> for
755 complete details.</CENTER>