ab(1) ab(1)
NAME
ab - Apache HTTP server benchmarking tool
SYNOPSIS
ab [ -k ] [ -e ] [ -q ] [ -S ] [ -i ] [ -s ] [ -n requests
] [ -t timelimit ] [ -c concurrency ] [ -p POST file ] [
-A Authenticate username:password ] [ -X proxy [ :port ] ]
[ -P Proxy Authenticate username:password ] [ -H Custom
header ] [ -C Cookie name=value ] [ -T content-type ] [ -v
verbosity ] ] [ -w output HTML ] ] [ -g output GNUPLOT ] ]
[ -e output CSV ] ] [ -x <table> attributes ] ] [ -y <tr>
attributes ] ] [ -z <td> attributes ] [http[s]://]host-
name[:port]/path
ab [ -V ] [ -h ]
DESCRIPTION
ab is a tool for benchmarking the performance of your
Apache HyperText Transfer Protocol (HTTP) server. It does
this by giving you an indication of how many requests per
second your Apache installation can serve.
OPTIONS
-k Enable the HTTP KeepAlive feature; that is,
perform multiple requests within one HTTP ses-
sion. Default is no KeepAlive.
-d Do not display the "percentage served within
XX [ms] table". (legacy support).
-S Do not display the median and standard devia-
tion values, nor display the warning/error
messages when the average and median are more
than one or two times the standard deviation
apart. And default to the min/avg/max values.
(legacy support).
-s When compiled in (bb -h will show you) use the
SSL protected https rather than the http pro-
tocol. This feature is experimental and very
rudimentary. You propably do not want to use
it.
-k Enable the HTTP KeepAlive feature; that is,
perform multiple requests within one HTTP ses-
sion. Default is no KeepAlive. -i Use an HTTP
'HEAD' instead of the GET method. Cannot be
mixed with POST.
-n requests The number of requests to perform for the
benchmarking session. The default is to per-
form just one single request, which will not
give representative benchmarking results.
-t timelimit
The number of seconds to spend benchmarking.
Using this option automatically set the number
of requests for the benchmarking session to
50000. Use this to benchmark the server for a
fixed period of time. By default, there is no
timelimit.
-c concurrency
The number of simultaneous requests to per-
form. The default is to perform one HTTP
request at at time, that is, no concurrency.
-p POST file
A file containing data that the program will
send to the Apache server in any HTTP POST
requests.
-A Authorization username:password
Supply Basic Authentication credentials to the
server. The username and password are sepa-
rated by a single ':', and sent as uuencoded
data. The string is sent regardless of
whether the server needs it; that is, has sent
a 401 Authentication needed.
-X proxy[:port]
Route all requests through the proxy (at
optional port).
-P Proxy-Authorization username:password
Supply Basic Authentication credentials to a
proxy en-route. The username and password are
separated by a single ':', and sent as uuen-
coded data. The string is sent regardless of
whether the proxy needs it; that is, has sent
a 407 Proxy authentication needed.
-C Cookie name=value
Add a 'Cookie:' line to the request. The argu-
ment is typically a 'name=value' pair. This
option may be repeated.
-p Header string
Append extra headers to the request. The argu-
ment is typically in the form of a valid
header line, usually a colon separated field
value pair, for example, 'Accept-Encoding:
zip/zop;8bit'.
-T content-type
The content-type header to use for POST data.
-g gnuplot file
Write all measured values out as a 'gnuplot'
or TSV (Tab separate values) file. This file
can easily be imported into packages like Gnu-
plot, IDL, Mathematica, Igor or even Excell.
The labels are on the first line of the file.
-q When processing more than 150 requsts; ab out-
puts a progress count on stderr every 10% or
100 requests or so. The -q flag qill suppress
these messages.
-e CSV file Write a Comma separated value (CSV) file which
contains for each percentage (from 1% to 100%)
the time (in milli seconds) it took to serve
that percentage of the requests. This is usu-
ally more usefull than the 'gnuplot' file; as
the results are already
-v Sets the verbosity level. Level 4 and above
prints information on headers, level 3 and
above prints response codes (for example, 404,
200), and level 2 and above prints warnings
and informational messages.
-w Print out results in HTML tables. The default
table is two columns wide, with a white back-
ground.
-x attributes
The string to use as attributes for <table>.
Attributes are inserted <table here >
-y attributes
The string to use as attributes for <tr>.
-z attributes
The string to use as attributes for <td>.
-V Display the version number and exit.
-h Display usage information.
BUGS
There are various statically declared buffers of fixed
length. Combined with inefficient parsing of the command
line arguments, the response headers from the server, and
other external inputs, these buffers might overflow.
Ab does not implement HTTP/1.x fully; instead, it only
accepts some 'expected' forms of responses.
The rather heavy use of strstr(3) by the program may skew
performance results, since it uses significant CPU
resources. Make sure that performance limits are not hit
by ab before your server's limit is reached.
The HTML output is not as complete as the text output.
Up to version 1.3d ab has propably reported values way to
low for most measurements; as a single timeout (which is
usually in the order of seconds) will shift several thou-
sands of milli-second responses by a considerable factor.
This was further componded by a serious interger overrun
which would for realistic run's (i.e. those longer than a
few minutes) produce believable but totally bogus results.
Thanks to Sander Temme for solving this riddle.
SEE ALSO
httpd(8)
October 1999 ab(1)