6.5. Optimizing the MySQL Server
6.5.1. System Factors and Startup Parameter Tuning
We start with system-level factors, because some of these decisions must be made very early to achieve large performance gains. In other cases, a quick look at this section may suffice. However, it is always nice to have a sense of how much can be gained by changing factors that apply at this level.
The operating system to use is very important. To get the best use of multiple-CPU machines, you should use Solaris (because its threads implementation works well) or Linux (because the 2.4 and later kernels have good SMP support). Note that older Linux kernels have a 2GB filesize limit by default. If you have such a kernel and a need for files larger than 2GB, you should get the Large File Support (LFS) patch for the ext2 filesystem. Other filesystems such as ReiserFS and XFS do not have this 2GB limitation.
Before using MySQL in production, we advise you to test it on your intended platform.
If you have enough RAM, you could remove all swap devices. Some operating systems use a swap device in some contexts even if you have free memory.
Avoid external locking. Since MySQL 4.0, the default has been for external locking to be disabled on all systems. The --external-locking and --skip-external-locking options explicitly enable and disable external locking.
Note that disabling external locking does not affect MySQL's functionality as long as you run only one server. Just remember to take down the server (or lock and flush the relevant tables) before you run myisamchk. On some systems it is mandatory to disable external locking because it does not work, anyway.
The only case in which you cannot disable external locking is when you run multiple MySQL servers (not clients) on the same data, or if you run myisamchk to check (not repair) a table without telling the server to flush and lock the tables first. Note that using multiple MySQL servers to access the same data concurrently is generally not recommended, except when using MySQL Cluster.
The LOCK TABLES and UNLOCK TABLES statements use internal locking, so you can use them even if external locking is disabled.
6.5.2. Tuning Server Parameters
You can determine the default buffer sizes used by the mysqld server using this command:
shell> mysqld --verbose --help
This command produces a list of all mysqld options and configurable system variables. The output includes the default variable values and looks something like this:
[Pages 473 - 474]
(No default value)
(No default value)
(No default value)
(No default value)
If there is a mysqld server currently running, you can see the current values of its system variables by connecting to it and issuing this statement:
You can also see some statistical and status indicators for a running server by issuing this statement:
System variable and status information also can be obtained using mysqladmin:
shell> mysqladmin variables
shell> mysqladmin extended-status
For a full description for all system and status variables, see Section 4.2.2, "Server System Variables," and Section 4.2.4, "Server Status Variables."
MySQL uses algorithms that are very scalable, so you can usually run with very little memory. However, normally you get better performance by giving MySQL more memory.
When tuning a MySQL server, the two most important variables to configure are key_buffer_size and table_cache. You should first feel confident that you have these set appropriately before trying to change any other variables.
The following examples indicate some typical variable values for different runtime configurations.
If you have at least 256MB of memory and many tables and want maximum performance with a moderate number of clients, you should use something like this:
shell> mysqld_safe --key_buffer_size=64M --table_cache=256 \
--sort_buffer_size=4M --read_buffer_size=1M &
If you have only 128MB of memory and only a few tables, but you still do a lot of sorting, you can use something like this:
shell> mysqld_safe --key_buffer_size=16M --sort_buffer_size=1M
If there are very many simultaneous connections, swapping problems may occur unless mysqld has been configured to use very little memory for each connection. mysqld performs better if you have enough memory for all connections.
With little memory and lots of connections, use something like this:
shell> mysqld_safe --key_buffer_size=512K --sort_buffer_size=100K \
Or even this:
shell> mysqld_safe --key_buffer_size=512K --sort_buffer_size=16K \
--table_cache=32 --read_buffer_size=8K \
If you are performing GROUP BY or ORDER BY operations on tables that are much larger than your available memory, you should increase the value of read_rnd_buffer_size to speed up the reading of rows following sorting operations.
When you have installed MySQL, the support-files directory contains some different my.cnf sample files: my-huge.cnf, my-large.cnf, my-medium.cnf, and my-small.cnf. You can use these as a basis for optimizing your system. (On Windows, look in the MySQL installation directory.)
If you specify an option on the command line for mysqld or mysqld_safe, it remains in effect only for that invocation of the server. To use the option every time the server runs, put it in an option file.
To see the effects of a parameter change, do something like this:
shell> mysqld --key_buffer_size=32M --verbose --help
The variable values are listed near the end of the output. Make sure that the --verbose and --help options are last. Otherwise, the effect of any options listed after them on the command line are not reflected in the output.
For information on tuning the InnoDB storage engine, see Section 8.2.11, "InnoDB Performance Tuning Tips."
6.5.3. Controlling Query Optimizer Performance
The task of the query optimizer is to find an optimal plan for executing an SQL query. Because the difference in performance between "good" and "bad" plans can be orders of magnitude (that is, seconds versus hours or even days), most query optimizers, including that of MySQL, perform a more or less exhaustive search for an optimal plan among all possible query evaluation plans. For join queries, the number of possible plans investigated by the MySQL optimizer grows exponentially with the number of tables referenced in a query. For small numbers of tables (typically less than 710) this is not a problem. However, when bigger queries are submitted, the time spent in query optimization may easily become the major bottleneck in the server's performance.
MySQL 5.0.1 introduces a more flexible method for query optimization that allows the user to control how exhaustive the optimizer is in its search for an optimal query evaluation plan. The general idea is that the fewer plans that are investigated by the optimizer, the less time it spends in compiling a query. On the other hand, because the optimizer skips some plans, it may miss finding an optimal plan.
The behavior of the optimizer with respect to the number of plans it evaluates can be controlled via two system variables:
The optimizer_prune_level variable tells the optimizer to skip certain plans based on estimates of the number of rows accessed for each table. Our experience shows that this kind of "educated guess" rarely misses optimal plans, and may dramatically reduce query compilation times. That is why this option is on (optimizer_prune_level=1) by default. However, if you believe that the optimizer missed a better query plan, this option can be switched off (optimizer_prune_level=0) with the risk that query compilation may take much longer. Note that, even with the use of this heuristic, the optimizer still explores a roughly exponential number of plans.
The optimizer_search_depth variable tells how far into the "future" of each incomplete plan the optimizer should look to evaluate whether it should be expanded further. Smaller values of optimizer_search_depth may result in orders of magnitude smaller query compilation times. For example, queries with 12, 13, or more tables may easily require hours and even days to compile if optimizer_search_depth is close to the number of tables in the query. At the same time, if compiled with optimizer_search_depth equal to 3 or 4, the optimizer may compile in less than a minute for the same query. If you are unsure of what a reasonable value is for optimizer_search_depth, this variable can be set to 0 to tell the optimizer to determine the value automatically.
6.5.4. How Compiling and Linking Affects the Speed of MySQL
Most of the following tests were performed on Linux with the MySQL benchmarks, but they should give some indication for other operating systems and workloads.
You obtain the fastest executables when you link with -static.
On Linux, it is best to compile the server with pgcc and -O3. You need about 200MB memory to compile sql_yacc.cc with these options, because gcc or pgcc needs a great deal of memory to make all functions inline. You should also set CXX=gcc when configuring MySQL to avoid inclusion of the libstdc++ library, which is not needed. Note that with some versions of pgcc, the resulting binary runs only on true Pentium processors, even if you use the compiler option indicating that you want the resulting code to work on all x586-type processors (such as AMD).
By using a better compiler and compilation options, you can obtain a 1030% speed increase in applications. This is particularly important if you compile the MySQL server yourself.
When we tested both the Cygnus CodeFusion and Fujitsu compilers, neither was sufficiently bug-free to allow MySQL to be compiled with optimizations enabled.
The standard MySQL binary distributions are compiled with support for all character sets. When you compile MySQL yourself, you should include support only for the character sets that you are going to use. This is controlled by the --with-charset option to configure.
Here is a list of some measurements that we have made:
If you use pgcc and compile everything with -O6, the mysqld server is 1% faster than with gcc 2.95.2.
If you link dynamically (without -static), the result is 13% slower on Linux. Note that you still can use a dynamically linked MySQL library for your client applications. It is the server that is most critical for performance.
For a connection from a client to a server running on the same host, if you connect using TCP/IP rather than a Unix socket file, performance is 7.5% slower. (On Unix, if you connect to the hostname localhost, MySQL uses a socket file by default.)
For TCP/IP connections from a client to a server, connecting to a remote server on another host is 811% slower than connecting to a server on the same host, even for connections over 100Mb/s Ethernet.
When running our benchmark tests using secure connections (all data encrypted with internal SSL support) performance was 55% slower than with unencrypted connections.
If you compile with --with-debug=full, most queries are 20% slower. Some queries may take substantially longer; for example, the MySQL benchmarks run 35% slower. If you use --with-debug (without =full), the speed decrease is only 15%. For a version of mysqld that has been compiled with --with-debug=full, you can disable memory checking at runtime by starting it with the --skip-safemalloc option. The execution speed should then be close to that obtained when configuring with --with-debug.
On a Sun UltraSPARC-IIe, a server compiled with Forte 5.0 is 4% faster than one compiled with gcc 3.2.
On a Sun UltraSPARC-IIe, a server compiled with Forte 5.0 is 4% faster in 32-bit mode than in 64-bit mode.
Compiling with gcc 2.95.2 for UltraSPARC with the -mcpu=v8 -Wa,-xarch=v8plusa options gives 4% more performance.
On Solaris 2.5.1, MIT-pthreads is 812% slower than Solaris native threads on a single processor. With greater loads or more CPUs, the difference should be larger.
Compiling on Linux-x86 using gcc without frame pointers (-fomit-frame-pointer or -fomit-frame-pointer -ffixed-ebp) makes mysqld 14% faster.
Binary MySQL distributions for Linux that are provided by MySQL AB used to be compiled with pgcc. We had to go back to regular gcc due to a bug in pgcc that would generate binaries that do not run on AMD. We will continue using gcc until that bug is resolved. In the meantime, if you have a non-AMD machine, you can build a faster binary by compiling with pgcc. The standard MySQL Linux binary is linked statically to make it faster and more portable.
6.5.5. How MySQL Uses Memory
The following list indicates some of the ways that the mysqld server uses memory. Where applicable, the name of the system variable relevant to the memory use is given:
The key buffer (variable key_buffer_size) is shared by all threads; other buffers used by the server are allocated as needed. See Section 6.5.2, "Tuning Server Parameters."
Each connection uses some thread-specific space:
A stack (default 192KB, variable thread_stack)
A connection buffer (variable net_buffer_length)
A result buffer (variable net_buffer_length)
The connection buffer and result buffer are dynamically enlarged up to max_allowed_ packet when needed. While a query is running, a copy of the current query string is also allocated.
All threads share the same base memory.
When a thread is no longer needed, the memory allocated to it is released and returned to the system unless the thread goes back into the thread cache. In that case, the memory remains allocated.
Only compressed MyISAM tables are memory mapped. This is because the 32-bit memory space of 4GB is not large enough for most big tables. When systems with a 64-bit address space become more common, we may add general support for memory mapping.
Each request that performs a sequential scan of a table allocates a read buffer (variable read_buffer_size).
When reading rows in an arbitrary sequence (for example, following a sort), a random-read buffer (variable read_rnd_buffer_size) may be allocated in order to avoid disk seeks.
All joins are executed in a single pass, and most joins can be done without even using a temporary table. Most temporary tables are memory-based hash tables. Temporary tables with a large row length (calculated as the sum of all column lengths) or that contain BLOB columns are stored on disk.
If an internal heap table exceeds the size of tmp_table_size, MySQL handles this automatically by changing the in-memory heap table to a disk-based MyISAM table as necessary. You can also increase the temporary table size by setting the tmp_table_size option to mysqld, or by setting the SQL option SQL_BIG_TABLES in the client program.
Most requests that perform a sort allocate a sort buffer and zero to two temporary files depending on the result set size.
Almost all parsing and calculating is done in a local memory store. No memory overhead is needed for small items, so the normal slow memory allocation and freeing is avoided. Memory is allocated only for unexpectedly large strings. This is done with malloc() and free().
For each MyISAM table that is opened, the index file is opened once; the data file is opened once for each concurrently running thread. For each concurrent thread, a table structure, column structures for each column, and a buffer of size 3 x N are allocated (where N is the maximum row length, not counting BLOB columns). A BLOB column requires five to eight bytes plus the length of the BLOB data. The MyISAM storage engine maintains one extra row buffer for internal use.
For each table having BLOB columns, a buffer is enlarged dynamically to read in larger BLOB values. If you scan a table, a buffer as large as the largest BLOB value is allocated.
Handler structures for all in-use tables are saved in a cache and managed as a FIFO. By default, the cache has 64 entries. If a table has been used by two running threads at the same time, the cache contains two entries for the table. See Section 6.4.8, "How MySQL Opens and Closes Tables."
A FLUSH TABLES statement or mysqladmin flush-tables command closes all tables that are not in use at once and marks all in-use tables to be closed when the currently executing thread finishes. This effectively frees most in-use memory. FLUSH TABLES does not return until all tables have been closed.
ps and other system status programs may report that mysqld uses a lot of memory. This may be caused by thread stacks on different memory addresses. For example, the Solaris version of ps counts the unused memory between stacks as used memory. You can verify this by checking available swap with swap -s. We test mysqld with several memory-leakage detectors (both commercial and Open Source), so there should be no memory leaks.
6.5.6. How MySQL Uses DNS
When a new client connects to mysqld, mysqld spawns a new thread to handle the request. This thread first checks whether the hostname is in the hostname cache. If not, the thread attempts to resolve the hostname:
If the operating system supports the thread-safe gethostbyaddr_r() and gethostbyname_r() calls, the thread uses them to perform hostname resolution.
If the operating system does not support the thread-safe calls, the thread locks a mutex and calls gethostbyaddr() and gethostbyname() instead. In this case, no other thread can resolve hostnames that are not in the hostname cache until the first thread unlocks the mutex.
You can disable DNS hostname lookups by starting mysqld with the --skip-name-resolve option. However, in this case, you can use only IP numbers in the MySQL grant tables.
If you have a very slow DNS and many hosts, you can get more performance by either disabling DNS lookups with --skip-name-resolve or by increasing the HOST_CACHE_SIZE define (default value: 128) and recompiling mysqld.
You can disable the hostname cache by starting the server with the --skip-host-cache option. To clear the hostname cache, issue a FLUSH HOSTS statement or execute the mysqladmin flush-hosts command.
To disallow TCP/IP connections entirely, start mysqld with the --skip-networking option.