Files
sysbench/doc/manual.xml
Alexey Kopytov d224c6368a Initial import
2006-01-20 16:13:23 +00:00

752 lines
34 KiB
XML

<?xml version="1.0" encoding='ISO-8859-1'?>
<!DOCTYPE book PUBLIC "-//OASIS//DTD DocBook XML V4.1.2//EN" "http://www.oasis-open.org/docbook/xml/4.1.2/docbookx.dtd">
<book id="sysbench-guide">
<bookinfo>
<title>SysBench manual</title>
<authorgroup>
<author>
<firstname>Alexey</firstname>
<surname>Kopytov</surname>
<affiliation>
<address><email>kaamos@users.sourceforge.net</email></address>
</affiliation>
</author>
</authorgroup>
<copyright>
<year>2004-2005</year>
<holder>MySQL AB</holder>
</copyright></bookinfo>
<toc></toc>
<chapter id="introduction">
<title>Introduction</title>
<para>
SysBench is a modular, cross-platform and multi-threaded benchmark tool for evaluating OS parameters
that are important for a system running a database under intensive load.
</para>
<para>
The idea of this benchmark suite is to quickly get an impression about system performance without setting up
complex database benchmarks or even without installing a database at all.
</para>
<section id="features">
<title>Features of SysBench</title>
<para>
Current features allow to test the following system parameters:
<itemizedlist>
<listitem><para>file I/O performance</para></listitem>
<listitem><para>scheduler performance</para></listitem>
<listitem><para>memory allocation and transfer speed</para></listitem>
<listitem><para>POSIX threads implementation performance</para></listitem>
<listitem><para>database server performance</para></listitem>
</itemizedlist>
</para>
</section>
<section id="design">
<title>Design</title>
<para>
The design is very simple. SysBench runs a specified number of threads and they all execute requests in parallel.
The actual workload produced by requests depends on the specified test mode. You can limit either the total number of
requests or the total time for the benchmark, or both.
</para>
<para>
Available test modes are implemented by compiled-in modules, and SysBench was designed to make adding new test modes
an easy task. Each test mode may have additional (or workload-specific) options.
</para>
</section>
<section id="links">
<title>Links</title>
<variablelist>
<varlistentry>
<term>Home page</term>
<listitem><para>
<ulink url="http://sysbench.sf.net/">http://sysbench.sf.net/</ulink>.
</para></listitem>
</varlistentry>
<varlistentry>
<term>Download</term>
<listitem><para>
<ulink url="http://sf.net/projects/sysbench/">http://sf.net/projects/sysbench/</ulink>.
</para></listitem>
</varlistentry>
<varlistentry>
<term>Mailing lists</term>
<listitem><para>
<ulink url="http://sourceforge.net/mail/?group_id=102348">sysbench-general</ulink>
</para></listitem>
</varlistentry>
<varlistentry>
<term>Web forums</term>
<listitem><para>
<itemizedlist>
<listitem>
<ulink url="http://sourceforge.net/forum/forum.php?forum_id=353125">Developers</ulink>
</listitem>
<listitem>
<ulink url="http://sourceforge.net/forum/forum.php?forum_id=353124">Help</ulink>
</listitem>
<listitem>
<ulink url="http://sourceforge.net/forum/forum.php?forum_id=353123">Open discussion</ulink>
</listitem>
</itemizedlist>
</para></listitem>
</varlistentry>
<varlistentry>
<term>Bug tracking system</term>
<listitem><para>
<itemizedlist>
<listitem>
<ulink url="http://sourceforge.net/tracker/?atid=631676&amp;group_id=102348&amp;func=browse">
Bug reports
</ulink>
</listitem>
<listitem>
<ulink url="http://sourceforge.net/tracker/?atid=631679&amp;group_id=102348&amp;func=browse">
Feature requests
</ulink>
</listitem>
</itemizedlist>
</para></listitem>
</varlistentry>
</variablelist>
</section>
<section id="install">
<title>Installation</title>
<para>
The following standart procedure will be sufficient to build SysBench in most cases:
<synopsis>
<command>./configure</command>
<command>make</command>
<command>make install</command>
</synopsis>
The above procedure will try to compile SysBench with MySQL support by default. If you have MySQL headers and libraries
in non-standard locations (and no <command>mysql_config</command> can be found in the <constant>PATH</constant>
environmental variable), then you can specify them explicitly with <option>--with-mysql-includes</option> and
<option>--with-mysql-libs</option> options to <command>./configure</command>.
</para>
<para>
To compile SysBench without MySQL support, use <option>--without-mysql</option>. In this case all database-related
test modes will be unavailable.
</para>
<para>
If you are running on a 64-bit platform, make sure to build a 64-bit binary by passing the proper target platform and compiler options to <command>configure</command> script. You can also consult the INSTALL file for generic installation instructions.
</para>
</section>
</chapter>
<chapter id="usage">
<title>Usage</title>
<section id="general_syntax">
<title>General syntax</title>
<para>
The general syntax for SysBench is as follows:
<screen>
sysbench <emphasis>[common-options]</emphasis> --test=<emphasis>name</emphasis> <emphasis>[test-options]</emphasis> <emphasis>command</emphasis>
</screen>
See <xref linkend="common_options"/> for a description of common options and documentation for particular test mode for a list
of test-specific options.
</para>
<para>
Below is a brief description of available commands and their purpose:
<variablelist>
<varlistentry>
<term><command>prepare</command></term>
<listitem>Performs preparative actions for those tests which need them, e.g. creating the necessary
files on disk for the <option>fileio</option> test, or filling the test database for the
<option>oltp</option> test.
</listitem>
</varlistentry>
<varlistentry>
<term><command>run</command></term>
<listitem>Runs the actual test specified with the <option>--test=<emphasis>name</emphasis></option> option.
</listitem>
</varlistentry>
<varlistentry>
<term><command>cleanup</command></term>
<listitem>Removes temporary data after the test run in those tests which create one.
</listitem>
</varlistentry>
<varlistentry>
<term><command>help</command></term>
<listitem>Displays usage information for a test specified with the
<option>--test=<emphasis>name</emphasis></option> option.
</listitem>
</varlistentry>
</variablelist>
</para>
<para>
Also you can use <command>sysbench help</command> to display the brief usage summary and the list of available test modes.
</para>
</section>
<section id="common_options">
<title>Common command line options</title>
<para>
The table below lists the supported common options, their descriptions and default values:
</para>
<informaltable frame="all">
<tgroup cols='3'>
<tbody>
<row><entry><emphasis>Option</emphasis></entry><entry><emphasis>Description</emphasis></entry><entry><emphasis>Default value</emphasis></entry></row>
<row><entry><option>--num-threads</option></entry><entry>The total number of worker threads to create</entry><entry>1</entry></row>
<row><entry><option>--max-requests</option></entry><entry>Limit for total number of requests. 0 means unlimited</entry><entry>10000</entry></row>
<row><entry><option>--max-time</option></entry><entry>Limit for total execution time in seconds. 0 (default) means unlimited</entry><entry>0</entry></row>
<row><entry><option>--thread-stack-size</option></entry><entry>Size of stack for each thread</entry><entry>32K</entry></row>
<row><entry><option>--init-rnd</option></entry><entry>Specifies if random numbers generator should be initialized from timer before the test start</entry><entry>off</entry></row>
<row><entry><option>--test</option></entry><entry>Name of the test mode to run</entry><entry><emphasis>Required</emphasis></entry></row>
<row><entry><option>--debug</option></entry><entry>Print more debug info</entry><entry>off</entry></row>
<row><entry><option>--validate</option></entry><entry>Perform
validation of test results where possible </entry><entry>off</entry></row>
<row><entry><option>--help</option></entry><entry>Print help on general syntax or on a test mode specified with --test, and exit</entry><entry>off</entry></row>
<row><entry><option>--verbosity</option></entry><entry>Verbosity level (0 - only critical messages, 5 - debug)</entry><entry>4</entry></row>
<row><entry><option>--percentile</option></entry>
<entry>
<para>
SysBench measures execution times for all processed
requests to display statistical information like minimal, average and
maximum execution time. For most benchmarks it is also useful to know
a request execution time value matching some percentile (e.g. 95%
percentile means we should drop 5% of the most long requests and
choose the maximal value from the remaining ones).
</para>
<para>
This option allows to specify a percentile rank of query
execution times to count
</para>
</entry>
<entry>95</entry></row>
<row><entry><option>--batch</option></entry><entry>Dump current results periodically (see <xref linkend="batch_mode"/>)</entry><entry>off</entry></row>
<row><entry><option>--batch-delay</option></entry><entry>Delay between batch dumps in secods (see <xref linkend="batch_mode"/>)</entry><entry>300</entry></row>
<row><entry><option>--validate</option></entry><entry>Perform validation of test results where possible</entry><entry>off</entry></row>
</tbody>
</tgroup>
</informaltable>
<para>
Note that numerical values for all <emphasis>size</emphasis> options
(like <option>--thread-stack-size</option> in this table) may be
specified by appending the corresponding multiplicative suffix (K for
kilobytes, M for megabytes, G for gigabytes and T for terabytes).
</para>
</section>
<section id="batch_mode">
<title>Batch mode</title>
In some cases it is useful to have not only the final benchmarks
statistics, but also periodical dumps of current stats to see how they
change over the test run. For this purpose SysBench has a batch
execution mode which is turned on by the <option>--batch</option>
option. You may specify the delay in seconds between the consequent dumps with
the <option>--batch-delay</option> option. Example:
<screen>
sysbench --batch --batch-delay=5 --test=threads run
</screen>
This will run SysBench in a threads test mode, with the current values
of minimum, average, maximum and percentile for request execution
times printed every 5 seconds.
</section>
<section id="test_modes">
<title>Test modes</title>
This section gives a detailed description for each test mode available
in SysBench.
<section id="cpu_mode">
<title><option>cpu</option></title>
</section>
<para>
The <option>cpu</option> is one of the most simple benchmarks in
SysBench. In this mode each request consists in calculation of prime numbers up to a value
specified by the <option>--cpu-max-primes</option> option. All calculations are performed using 64-bit integers.
</para>
<para>
Each thread executes the requests concurrently until either the total number of requests or the total execution
time exceed the limits specified with the common command line options.
</para>
<para>
Example:
<screen>
sysbench --test=cpu --cpu-max-prime=20000 run
</screen>
</para>
<section id="threads_mode">
<title><option>threads</option></title>
</section>
<para>
This test mode was written to benchmark scheduler performance, more specifically the cases
when a scheduler has a large number of threads competing for some set of mutexes.
</para>
<para>
SysBench creates a specified number of threads and a specified number of mutexes. Then each thread
starts running the requests consisting of locking the mutex, yielding the CPU, so the thread is
placed in the run queue by the scheduler, then unlocking the mutex when the thread is rescheduled back
to execution. For each request, the above actions are run several times in a loop, so the more iterations
is performed, the more concurrency is placed on each mutex.
</para>
<para>
The following options are available in this test mode:
<informaltable frame="all">
<tgroup cols='3'>
<tbody>
<row><entry><emphasis>Option</emphasis></entry><entry><emphasis>Description</emphasis></entry><entry><emphasis>Default value</emphasis></entry></row>
<row><entry><option>--thread-yields</option></entry><entry>Number of <emphasis>lock/yield/unlock</emphasis> loops to execute per each request</entry><entry>1000</entry></row>
<row><entry><option>--thread-locks</option></entry><entry>Number of mutexes to create</entry><entry>8</entry></row>
</tbody>
</tgroup>
</informaltable>
</para>
<para>
Example:
<screen>
sysbench --num-threads=64 --test=threads --thread-yields=100 --thread-locks=2 run
</screen>
</para>
<section id="mutex_mode">
<title><option>mutex</option></title>
</section>
<para>
This test mode was written to emulate a situation when all threads run concurrently most of the time,
acquiring the mutex lock only for a short period of time (incrementing a global variable). So the purpose
of this benchmarks is to examine the performance of mutex implementation.
</para>
<para>
The following options are available in this test mode:
<informaltable frame="all">
<tgroup cols='3'>
<tbody>
<row><entry><emphasis>Option</emphasis></entry><entry><emphasis>Description</emphasis></entry><entry><emphasis>Default value</emphasis></entry></row>
<row><entry><option>--mutex-num</option></entry><entry>Number of mutexes. The actual mutex to lock is chosen randomly before each lock</entry><entry>4096</entry></row>
<row><entry><option>--mutex-locks</option></entry><entry>Number of mutex locks to acquire per each request</entry><entry>50000</entry></row>
<row><entry><option>--mutex-loops</option></entry><entry>Number of iterations for an empty loop to perform before acquiring the lock</entry><entry>10000</entry></row>
</tbody>
</tgroup>
</informaltable>
</para>
<section id="memory_mode">
<title><option>memory</option></title>
<para>
This test mode can be used to benchmark sequential memory reads or writes. Depending on command line
options each thread can access either a global or a local block for all memory operations.
</para>
<para>
The following options are available in this test mode:
<informaltable frame="all">
<tgroup cols='3'>
<tbody>
<row><entry><emphasis>Option</emphasis></entry><entry><emphasis>Description</emphasis></entry><entry><emphasis>Default value</emphasis></entry></row>
<row><entry><option>--memory-block-size</option></entry><entry>Size of memory block to use</entry><entry>1K</entry></row>
<row><entry><option>--memory-scope</option></entry><entry>
Possible values: <option>global</option>, <option>local</option>. Specifies whether each thread will
use a globally allocated memory block, or a local one.
</entry><entry>global</entry></row>
<row><entry><option>--memory-total-size</option></entry><entry>Total size of data to transfer</entry><entry>100G</entry></row>
<row><entry><option>--memory-oper</option></entry><entry>
Type of memory operations. Possible values: <option>read</option>, <option>write</option>.
</entry><entry>100G</entry></row>
</tbody>
</tgroup>
</informaltable>
</para>
</section>
<section id="fileio_mode">
<title><option>fileio</option></title>
<para>
This test mode can be used to produce various kinds of file I/O workloads. At the <option>prepare</option>
stage SysBench creates a specified number of files with a specified total size, then at the <option>run</option>
stage, each thread performs specified I/O operations on this set of files.
</para>
<para>
When the global <option>--validate</option> option is used with the <option>fileio</option> test mode, SysBench
performs checksums validation on all data read from the disk. On each write operation the block is filled with random values,
then the checksum is calculated
and stored in the block along with the offset of this block within a file. On each read operation the block is validated
by comparing the stored offset with the real offset, and the stored checksum with the real calculated checksum.
</para>
<para>
The following I/O operations are supported:
<variablelist>
<varlistentry>
<term><command>seqwr</command></term>
<listitem>sequential write
</listitem>
</varlistentry>
<varlistentry>
<term><command>seqrewr</command></term>
<listitem>sequential rewrite
</listitem>
</varlistentry>
<varlistentry>
<term><command>seqrd</command></term>
<listitem>sequential read
</listitem>
</varlistentry>
<varlistentry>
<term><command>rndrd</command></term>
<listitem>random read
</listitem>
</varlistentry>
<varlistentry>
<term><command>rndwr</command></term>
<listitem>random write
</listitem>
</varlistentry>
<varlistentry>
<term><command>rndrw</command></term>
<listitem>combined random read/write
</listitem>
</varlistentry>
</variablelist>
</para>
<para>
Also, the following file access modes can be specified, if the underlying platform supports them:
<variablelist>
<varlistentry>
<term>Asynchronous I/O mode</term>
<listitem>
At the moment only Linux AIO implementation is supported. When running in asynchronous mode,
SysBench queues a specified number of I/O requests using Linux AIO API, then waits for
at least one of submitted requests to complete. After that a new series of I/O requests
is submitted.
</listitem>
</varlistentry>
<varlistentry>
<term>Slow <option>mmap()</option> mode</term>
<listitem>In this mode SysBench will use <option>mmap</option>'ed I/O. However, a separate
<option>mmap</option> will be used for each I/O request due to the limitation of 32-bit
architectures (we cannot <option>mmap()</option> the whole file, as its size migth possibly
exceed the maximum of 2 GB of the process address space).
</listitem>
</varlistentry>
<varlistentry>
<term>Fast <option>mmap()</option> mode</term>
<listitem>On 64-bit architectures it is possible to <option>mmap()</option> the whole file
into the process address space, avoiding the limitation of 2 GB on 32-bit platforms.
</listitem>
</varlistentry>
<varlistentry>
<term>Using <option>fdatasync()</option> instead of <option>fsync()</option></term>
Flush only data buffers, but not the metadata.
<listitem>
</listitem>
</varlistentry>
<varlistentry>
<term>Additional flags to <option>open(2)</option></term>
<listitem>SysBench can use additional flags to <option>open(2)</option>, such as <option>O_SYNC</option>,
<option>O_DSYNC</option> and <option>O_DIRECT</option>.
</listitem>
</varlistentry>
</variablelist>
</para>
<para>
Below is a list of test-specific option for the <command>fileio</command> mode:
<informaltable frame="all">
<tgroup cols='3'>
<tbody>
<row><entry><emphasis>Option</emphasis></entry><entry><emphasis>Description</emphasis></entry><entry><emphasis>Default value</emphasis></entry></row>
<row><entry><option>--file-num</option></entry><entry>Number of files to create</entry><entry>128</entry></row>
<row><entry><option>--file-block-size</option></entry><entry>
Block size to use in all I/O operations
</entry><entry>16K</entry></row>
<row><entry><option>--file-total-size</option></entry><entry>Total size of files</entry><entry>2G</entry></row>
<row><entry><option>--file-test-mode</option></entry><entry>
Type of workload to produce. Possible values: <option>seqwr</option>, <option>seqrewr</option>,
<option>seqrd</option>, <option>rndrd</option>, <option>rndwr</option>, <option>rndwr</option> (see above)
</entry><entry><emphasis>required</emphasis></entry></row>
<row><entry><option>--file-io-mode</option></entry><entry>
I/O mode. Possible values: <option>sync</option>, <option>async</option>, <option>fastmmap</option>,
<option>slowmmap</option> (only if supported by the platform, see above).
</entry><entry>sync</entry></row>
<row><entry><option>--file-async-backlog</option></entry><entry>
Number of asynchronous operations to queue per thread (only for <option>--file-io-mode=async</option>, see above)
</entry><entry>128</entry></row>
<row><entry><option> --file-extra-flags</option></entry><entry>
Additional flags to use with <option>open(2)</option>
</entry><entry></entry></row>
<row><entry><option>--file-fsync-freq</option></entry><entry>
Do <option>fsync()</option> after this number of requests (0 - don't use <option>fsync()</option>)
</entry><entry>100</entry></row>
<row><entry><option>--file-fsync-all</option></entry><entry>
Do <option>fsync()</option> after each write operation
</entry><entry>no</entry></row>
<row><entry><option>--file-fsync-end</option></entry><entry>
Do <option>fsync()</option> at the end of the test
</entry><entry>yes</entry></row>
<row><entry><option>--file-fsync-mode</option></entry><entry>
Which method to use for synchronization. Possible values: <option>fsync</option>, <option>fdatasync</option> (see above)
</entry><entry>fsync</entry></row>
<row><entry><option>--file-merged-requests</option></entry><entry>
Merge at most this number of I/O requests if possible (0 - don't merge)
</entry><entry>0</entry></row>
<row><entry><option>--file-rw-ratio</option></entry><entry>
reads/writes ration for combined random read/write test
</entry><entry>1.5</entry></row>
</tbody>
</tgroup>
</informaltable>
</para>
<para>
Usage example:
<screen>
$ sysbench --num-threads=16 --test=fileio --file-total-size=3G --file-test-mode=rndrw prepare
$ sysbench --num-threads=16 --test=fileio --file-total-size=3G --file-test-mode=rndrw run
$ sysbench --num-threads=16 --test=fileio --file-total-size=3G --file-test-mode=rndrw cleanup
</screen>
In the above example the first command creates 128 files with the total size of 3 GB in the current directory, the
second command runs the actual benchmark and displays the results upon completion, and the third one removes the files
used for the test.
</para>
</section>
<section id="database_mode">
<title><option>oltp</option></title>
</section>
<para>
This test mode was written to benchmark a real database performance. At the <command>prepare</command> stage
the following table is created in the specified database (<option>sbtest</option> by default):
<screen>
CREATE TABLE `sbtest` (
`id` int(10) unsigned NOT NULL auto_increment,
`k` int(10) unsigned NOT NULL default '0',
`c` char(120) NOT NULL default '',
`pad` char(60) NOT NULL default '',
PRIMARY KEY (`id`),
KEY `k` (`k`);
</screen>
Then this table is filled with a specified number of rows.
</para>
<para>
The following execution modes are available at the <command>run</command> stage:
<variablelist>
<varlistentry>
<term>Simple</term>
<listitem>
<para>
In this mode each thread runs simple queries of the following form:
<screen> SELECT c FROM sbtest WHERE id=<emphasis>N</emphasis> </screen>
where <emphasis>N</emphasis> takes a random value in range 1..<emphasis>&lt;table size&gt;</emphasis>
</para>
</listitem>
</varlistentry>
<varlistentry>
<term>Advanced transactional</term>
<listitem>
<para>
Each thread performs transactions on the test table. If the test table and database support transactions
(e.g. InnoDB engine in MySQL), then <option>BEGIN</option>/<option>COMMIT</option> statements will be used
to start/stop a transaction. Otherwise, SysBench will use <option>LOCK TABLES</option>/<option>UNLOCK TABLES
</option> statements (e.g. for MyISAM engine in MySQL). If some rows are deleted in a transaction,
the same rows will be inserted within the same transaction, so this test mode does not destruct any data
in the test table and can be run multiple times on the same table.
</para>
Depending on the command line options, each transaction may contain the following statements:
<itemizedlist>
<listitem>Point queries: <screen>SELECT c FROM sbtest WHERE id=<emphasis>N</emphasis></screen></listitem>
<listitem>
Range queries: <screen>SELECT c FROM sbtest WHERE id BETWEEN <emphasis>N</emphasis> AND <emphasis>M</emphasis> </screen> </listitem>
<listitem>
Range SUM() queries: <screen>SELECT SUM(K) FROM sbtest WHERE id BETWEEN <emphasis>N</emphasis> and <emphasis>M</emphasis></screen></listitem>
<listitem>Range ORDER BY queries:<screen>SELECT c FROM sbtest WHERE id between <emphasis>N</emphasis> and <emphasis>M</emphasis> ORDER BY c</screen></listitem>
<listitem>Range DISTINCT queries:<screen>SELECT DISTINCT c FROM sbtest WHERE id BETWEEN <emphasis>N</emphasis> and <emphasis>M</emphasis> ORDER BY c</screen> </listitem>
<listitem>UPDATEs on index column:<screen>UPDATE sbtest SET k=k+1 WHERE id=<emphasis>N</emphasis> </screen> </listitem>
<listitem>UPDATEs on non-index column:<screen>UPDATE sbtest SET c=<emphasis>N</emphasis> WHERE id=<emphasis>M</emphasis> </screen> </listitem>
<listitem>DELETE queries:<screen>DELETE FROM sbtest WHERE id=<emphasis>N</emphasis> </screen> </listitem>
<listitem>INSERT queries:<screen>INSERT INTO sbtest VALUES (<emphasis>...</emphasis>) </screen> </listitem>
</itemizedlist>
</listitem>
</varlistentry>
<varlistentry>
<term>Non-transactional</term>
<listitem>
<para>
This mode is similar to <command>Simple</command>, but you can also choose the query to run. Note that unlike the
<command>Advanced transactional</command> mode, this one does not preserve the test table between requests, so
you should recreate it with the appropriate <command>cleanup</command>/<command>prepare</command> commands between
consecutive benchmarks.
</para>
<para>
Below is a list of possible queries:
<itemizedlist>
<listitem>
Point queries:
<screen>SELECT pad FROM sbtest WHERE id=<emphasis>N</emphasis></screen>
</listitem>
<listitem>
UPDATEs on index column:
<screen>UPDATE sbtest SET k=k+1 WHERE id=<emphasis>N</emphasis></screen>
</listitem>
<listitem>
UPDATEs on non-index column:
<screen>UPDATE sbtest SET c=<emphasis>N</emphasis> WHERE id=<emphasis>M</emphasis></screen>
</listitem>
<listitem>
DELETE queries:
<screen>DELETE FROM sbtest WHERE id=<emphasis>N</emphasis></screen>
The generated row IDs are unique over each test run, so no row is deleted twice.
</listitem>
<listitem>
INSERT queries:
<screen>INSERT INTO sbtest (k, c, pad) VALUES(<emphasis>N</emphasis>, <emphasis>M</emphasis>, <emphasis>S</emphasis>)</screen>
</listitem>
</itemizedlist>
</para>
</listitem>
</varlistentry>
</variablelist>
</para>
<para>
Below is a list of options available for the database test mode:
<informaltable frame="all">
<tgroup cols='3'>
<tbody>
<row><entry><emphasis>Option</emphasis></entry><entry><emphasis>Description</emphasis></entry><entry><emphasis>Default value</emphasis></entry></row>
<row><entry><option>--oltp-test-mode</option></entry><entry>Execution mode (see above). Possible values: <option>simpe</option> (simple), <option>complex</option> (advanced transactional) and <option>nontrx</option> (non-transactional)</entry><entry><option>complex</option></entry></row>
<row><entry><option>--oltp-read-only</option></entry><entry>
Read-only mode. No <option>UPDATE</option>, <option>DELETE</option> or <option>INSERT</option> queries will be performed.
</entry><entry>off</entry></row>
<row><entry><option>--oltp-range-size</option></entry><entry>Range size for range queries</entry><entry>100</entry></row>
<row><entry><option>--oltp-point-selects</option></entry><entry>
Number of point select queries in a single transaction
</entry><entry>10</entry></row>
<row><entry><option>--oltp-simple-ranges</option></entry><entry>
Number of simple range queries in a single transaction
</entry><entry>1</entry></row>
<row><entry><option>--oltp-sum-ranges</option></entry><entry>
Number of SUM range queries in a single transaction
</entry><entry>1</entry></row>
<row><entry><option>--oltp-order-ranges</option></entry><entry>
Number of ORDER range queries in a single transaction
</entry><entry>1</entry></row>
<row><entry><option>--oltp-distinct-ranges</option></entry><entry>
Number of DISTINCT range queries in a single transaction
</entry><entry>1</entry></row>
<row><entry><option>--oltp-index-updates</option></entry><entry>
Number of index UPDATE queries in a single transaction
</entry><entry>1</entry></row>
<row><entry><option>--oltp-non-index-updates</option></entry><entry>
Number of non-index UPDATE queries in a single transaction
</entry><entry>1</entry></row>
<row><entry><option> --oltp-nontrx-mode</option></entry><entry>
Type of queries for non-transactional execution mode (see above). Possible values: <option>select</option>,
<option>update_key</option>, <option>update_nokey</option>, <option>insert</option>, <option>delete</option>.
</entry><entry><option>select</option></entry></row>
<row><entry><option>--oltp-connect-delay</option></entry><entry>
Time in microseconds to sleep after each connection to database
</entry><entry>10000</entry></row>
<row><entry><option>--oltp-user-delay-min</option></entry><entry>
Minimum time in microseconds to sleep after each request
</entry><entry>0</entry></row>
<row><entry><option>--oltp-user-delay-max</option></entry><entry>
Maximum time in microseconds to sleep after each request
</entry><entry>0</entry></row>
<row><entry><option>--oltp-table-name</option></entry><entry>
Name of the test table
</entry><entry>sbtest</entry></row>
<row><entry><option>--oltp-table-size</option></entry><entry>
Number of rows in the test table
</entry><entry>10000</entry></row>
<row><entry><option>--oltp-dist-type</option></entry><entry>
<para>
Distribution of random numbers. Possible values: <option>uniform</option> (uniform distribution),
<option>gauss</option> (gaussian distribution) and <option>special</option>.
</para>
<para>
With special distribution a specified percent of numbers is generated in a specified percent of cases (see options below).
</para>
</entry><entry><option>special</option></entry></row>
<row><entry><option>--oltp-dist-pct</option></entry><entry>
Percentage of values to be treated as 'special' (for special distribution)
</entry><entry>1</entry></row>
<row><entry><option>--oltp-dist-res</option></entry><entry>
Percentage of cases when 'special' values are generated (for special distribution)
</entry><entry>75</entry></row>
<row><entry><option>--db-ps-mode</option></entry><entry>
If the database driver supports Prepared Statements API, SysBench will use server-side prepared statements for all queries
where possible. Otherwise, client-side (or emulated) prepared statements will be used. This option allows to force using
emulation even when PS API is available. Possible values: <option>disable</option>, <option>auto</option>.
</entry><entry><option>auto</option></entry></row>
</tbody>
</tgroup>
</informaltable>
</para>
<para>
Also, each database driver may provide its own options. Currently only MySQL driver is available. Below is a list of MySQL-specific options:
<informaltable frame="all">
<tgroup cols='3'>
<tbody>
<row><entry><emphasis>Option</emphasis></entry><entry><emphasis>Description</emphasis></entry><entry><emphasis>Default value</emphasis></entry></row>
<row><entry><option>--mysql-host</option></entry><entry>MySQL server host</entry><entry><option>localhost</option></entry></row>
<row><entry><option>--mysql-port</option></entry><entry>
MySQL server port (in case TCP/IP connection should be used)
</entry><entry>3306</entry></row>
<row><entry><option>--mysql-socket</option></entry><entry>Unix socket file to communicate with the MySQL server</entry><entry></entry></row>
<row><entry><option>--mysql-user</option></entry><entry>
MySQL user
</entry><entry>user</entry></row>
<row><entry><option>--mysql-password</option></entry><entry>
MySQL password
</entry><entry></entry></row>
<row><entry><option>--mysql-db</option></entry><entry>
MySQL database name. Note SysBench will not automatically create this database. You should create it manually and grant
the appropriate privileges to a user which will be used to access the test table.
</entry><entry>sbtest</entry></row>
<row><entry><option>--mysql-table-type</option></entry><entry>
Type of the test table. Possible values: <option>myisam</option>, <option>innodb</option>, <option>heap</option>,
<option>ndbcluster</option>, <option>bdb</option>.
</entry><entry>innodb</entry></row>
<row><entry><option>--myisam-max-rows</option></entry><entry>
MAX_ROWS option for MyISAM tables (required for big tables)
</entry><entry>1000000</entry></row>
</tbody>
</tgroup>
</informaltable>
</para>
<para>
Example usage:
<screen>
$ sysbench --test=oltp --mysql-table-type=myisam --oltp-table-size=1000000 --mysql-socket=/tmp/mysql.sock prepare
$ sysbench --num-threads=16 --max-requests=100000 --test=oltp --oltp-table-size=1000000 --mysql-socket=/tmp/mysql.sock --oltp-read-only run
</screen>
The first command will create a MyISAM table 'sbtest' in a database 'sbtest' on a MySQL server using <option>/tmp/mysql.sock</option> socket, then fill this table with 1M records. The second command will run the actual benchmark with 16 client threads, limiting the total number of request by 100,000.
</para>
</section>
</chapter>
</book>