I'm trying to optimize a simple sql query that will be run over a ton of data lot of times.
This is the scenario:
- MySQL with InnoDB tables
- All fields used in where and join are Indexed. Tables have FK's.
- I don't need a whole cache of the query, but a cache of each table it's possible.
- Tables have much more reads that updates / inserts.
A real query:
SELECT one_field, another_field
FROM big_table
INNER JOIN medium_table ON ( ... )
INNER JOIN small_table ON ( ... )
WHERE week >= #number AND week <= #number AND year >= #number AND year <= #number
AND medium_table.indexed_field = #number
AND small_table.pk= #number
Some test results:
- Average query time: 2,3374s
- Test tables contains the tenth part of data (or even less) that real tables, like 75.000 rows for big_test_table.
- Query returns (in test cases) 6500 rows.
- Navicat says that 99.95% of query time is spended on "Sending data"
- I've tried with MyISAM, and the average query time was: 2,1898s
How can I improve the query or the database scheme?
It's better concatenate my year / week fields and in a [yearweek] indexed field?
In some benchmarks I see that MyISAM is faster than Innodb, but seems I have a little improvement:
http://www.mysqlperformanceblog.com/files/benchmarks/innodb-myisam-falcon.html
http://www.jortk.nl/2008/12/mysql-performance-benchmark-myisam-versus-innodb/
Is it worth lose FKs and go to MyISAM?
Thanks!
Updates:
Explain of the select:
1 SIMPLE small_table const PRIMARY PRIMARY 4 const 1 Using index
1 SIMPLE medium_table index_merge PRIMARY,Fk_small_table, Fk_indexed_field Fk_small_table, Fk_indexed_field 5,5 250 Using intersect(Fk_small_table, Fk_indexed_field); Using where
1 SIMPLE big_table ref Fk_code_of_medium_table,week-year-compound-index Fk_three 5 medium_table.primary_key 4 Using where
DB schema:
CREATE TABLE IF NOT EXISTS `db`.`big_table` (
`primary_key` BIGINT UNSIGNED NOT NULL AUTO_INCREMENT ,
`code_of_medium_table` INT UNSIGNED NULL ,
`week` SMALLINT UNSIGNED NULL ,
`year` YEAR NULL ,
......
PRIMARY KEY (`primary_key`) ,
INDEX `Fk_code_of_medium_table` (`code_of_medium_table` ASC) ,
INDEX `In_week-year` (`week` ASC, `year` ASC) ,
CONSTRAINT `Fk_code_of_medium_table`
FOREIGN KEY (`code_of_medium_table` )
REFERENCES `db`.`medium_table` (`code` )
ON DELETE NO ACTION
ON UPDATE NO ACTION,
ENGINE = InnoDB
CREATE TABLE IF NOT EXISTS `db`.`medium_table` (
`primary_key` INT UNSIGNED NOT NULL AUTO_INCREMENT ,
.....
`code_of_small_table` INT UNSIGNED NULL ,
`indexed_field ` INT UNSIGNED NULL ,
PRIMARY KEY (`primary_key`) ,
INDEX `Fk_code_of_small_table` (`code_of_small_table` ASC) ,
INDEX `Fk_indexed_field` (`another_field` ASC) ,
CONSTRAINT `Fk_code_of_small_table`
FOREIGN KEY (`code_of_small_table` )
REFERENCES `db`.`small_table` (`code` )
ON DELETE NO ACTION
ON UPDATE NO ACTION,
CONSTRAINT `Fk_indexed_field `
FOREIGN KEY (`indexed_field ` )
REFERENCES `db`.`other_table` (`code` )
ON DELETE NO ACTION
ON UPDATE NO ACTION)
ENGINE = InnoDB
CREATE TABLE IF NOT EXISTS `db`.`small_table` (
`primary_key` INT UNSIGNED NOT NULL AUTO_INCREMENT ,
.....
PRIMARY KEY (`primary_key`) )
ENGINE = InnoDB
Variables output:
auto_increment_increment 1
auto_increment_offset 1
automatic_sp_privileges ON
back_log 50
basedir /usr/
bdb_cache_size 8384512
bdb_home /var/lib/mysql/
bdb_log_buffer_size 262144
bdb_logdir
bdb_max_lock 10000
bdb_shared_data OFF
bdb_tmpdir /tmp/
binlog_cache_size 32768
bulk_insert_buffer_size 8388608
character_set_client utf8
character_set_connection utf8
character_set_database latin1
character_set_filesystem binary
character_set_results utf8
character_set_server latin1
character_set_system utf8
character_sets_dir /usr/share/mysql/charsets/
collation_connection utf8_general_ci
collation_database latin1_swedish_ci
collation_server latin1_swedish_ci
completion_type 0
concurrent_insert 1
connect_timeout 10
datadir /var/lib/mysql/
date_format %Y-%m-%d
datetime_format %Y-%m-%d %H:%i:%s
default_week_format 0
delay_key_write ON
delayed_insert_limit 100
delayed_insert_timeout 300
delayed_queue_size 1000
div_precision_increment 4
keep_files_on_create OFF
engine_condition_pushdown OFF
expire_logs_days 0
flush OFF
flush_time 0
ft_boolean_syntax + -><()~*:""&|
ft_max_word_len 84
ft_min_word_len 4
ft_query_expansion_limit 20
ft_stopword_file (built-in)
group_concat_max_len 1024
have_archive NO
have_bdb YES
have_blackhole_engine NO
have_compress YES
have_crypt YES
have_csv NO
have_dynamic_loading YES
have_example_engine NO
have_federated_engine NO
have_geometry YES
have_innodb YES
have_isam NO
have_merge_engine YES
have_ndbcluster NO
have_openssl DISABLED
have_ssl DISABLED
have_query_cache YES
have_raid NO
have_rtree_keys YES
have_symlink YES
hostname EarnedBE002
init_connect
init_file
init_slave
innodb_additional_mem_pool_size 1048576
innodb_autoextend_increment 8
innodb_buffer_pool_awe_mem_mb 0
innodb_buffer_pool_size 8388608
innodb_checksums ON
innodb_commit_concurrency 0
innodb_concurrency_tickets 500
innodb_data_file_path ibdata1:10M:autoextend
innodb_data_home_dir
innodb_adaptive_hash_index ON
innodb_doublewrite ON
innodb_fast_shutdown 1
innodb_file_io_threads 4
innodb_file_per_table OFF
innodb_flush_log_at_trx_commit 1
innodb_flush_method
innodb_force_recovery 0
innodb_lock_wait_timeout 50
innodb_locks_unsafe_for_binlog OFF
innodb_log_arch_dir
innodb_log_archive OFF
innodb_log_buffer_size 1048576
innodb_log_file_size 5242880
innodb_log_files_in_group 2
innodb_log_group_home_dir ./
innodb_max_dirty_pages_pct 90
innodb_max_purge_lag 0
innodb_mirrored_log_groups 1
innodb_open_files 300
innodb_rollback_on_timeout OFF
innodb_support_xa ON
innodb_sync_spin_loops 20
innodb_table_locks ON
innodb_thread_concurrency 8
innodb_thread_sleep_delay 10000
interactive_timeout 28800
join_buffer_size 131072
key_buffer_size 8384512
key_cache_age_threshold 300
key_cache_block_size 1024
key_cache_division_limit 100
language /usr/share/mysql/english/
large_files_support ON
large_page_size 0
large_pages OFF
lc_time_names en_US
license GPL
local_infile ON
locked_in_memory OFF
log OFF
log_bin OFF
log_bin_trust_function_creators OFF
log_error
log_queries_not_using_indexes OFF
log_slave_updates OFF
log_slow_queries OFF
log_warnings 1
long_query_time 10
low_priority_updates OFF
lower_case_file_system OFF
lower_case_table_names 0
max_allowed_packet 1048576
max_binlog_cache_size 18446744073709547520
max_binlog_size 1073741824
max_connect_errors 10
max_connections 100
max_delayed_threads 20
max_error_count 64
max_heap_table_size 16777216
max_insert_delayed_threads 20
max_join_size 18446744073709551615
max_length_for_sort_data 1024
max_prepared_stmt_count 16382
max_relay_log_size 0
max_seeks_for_key 18446744073709551615
max_sort_length 1024
max_sp_recursion_depth 0
max_tmp_tables 32
max_user_connections 0
max_write_lock_count 18446744073709551615
multi_range_count 256
myisam_data_pointer_size 6
myisam_max_sort_file_size 9223372036853727232
myisam_recover_options OFF
myisam_repair_threads 1
myisam_sort_buffer_size 8388608
myisam_stats_method nulls_unequal
net_buffer_length 16384
net_read_timeout 30
net_retry_count 10
net_write_timeout 60
new OFF
old_passwords OFF
open_files_limit 1024
optimizer_prune_level 1
optimizer_search_depth 62
pid_file /var/run/mysqld/mysqld.pid
plugin_dir
port 3306
preload_buffer_size 32768
profiling OFF
profiling_history_size 15
protocol_version 10
query_alloc_block_size 8192
query_cache_limit 1048576
query_cache_min_res_unit 4096
query_cache_size 0
query_cache_type ON
query_cache_wlock_invalidate OFF
query_prealloc_size 8192
range_alloc_block_size 4096
read_buffer_size 131072
read_only OFF
read_rnd_buffer_size 262144
relay_log
relay_log_index
relay_log_info_file relay-log.info
relay_log_purge ON
relay_log_space_limit 0
rpl_recovery_rank 0
secure_auth OFF
secure_file_priv
server_id 0
skip_external_locking ON
skip_networking OFF
skip_show_database OFF
slave_compressed_protocol OFF
slave_load_tmpdir /tmp/
slave_net_timeout 3600
slave_skip_errors OFF
slave_transaction_retries 10
slow_launch_time 2
socket /var/lib/mysql/mysql.sock
sort_buffer_size 2097144
sql_big_selects ON
sql_mode
sql_notes ON
sql_warnings OFF
ssl_ca
ssl_capath
ssl_cert
ssl_cipher
ssl_key
storage_engine MyISAM
sync_binlog 0
sync_frm ON
system_time_zone UTC
table_cache 64
table_lock_wait_timeout 50
table_type MyISAM
thread_cache_size 0
thread_stack 262144
time_format %H:%i:%s
time_zone SYSTEM
timed_mutexes OFF
tmp_table_size 33554432
tmpdir /tmp/
transaction_alloc_block_size 8192
transaction_prealloc_size 4096
tx_isolation REPEATABLE-READ
updatable_views_with_limit YES
version 5.0.77
version_bdb Sleepycat Software: Berkeley DB 4.1.24: (January 29, 2009)
version_comment Source distribution
version_compile_machine x86_64
version_compile_os redhat-linux-gnu
wait_timeout 28800
With the index proposed by ypercube, at moment, this is the fastest query (2,107s avg):
SELECT SQL_NO_CACHE int_field, varchar_field
FROM big_table
INNER JOIN medium_table ON ( ... )
INNER JOIN small_table ON ( ... )
WHERE week BETWEEN #number AND #number
AND year BETWEEN #number AND #number
AND small_table.pk = 248
AND medium_table.indexed_field = #number
I hope you find the following links and previous answers of mine of interest.
I wish I could provide you with a specific answer to your problem but I dont have the time to get my head completely around your schema: naming conventions are so bad, lack of useful information, inconsistencies etc, etc...
However, you should find the solution to your problem below and it's nothing to do with switching to the less performant myisam engine !
A snippet from one of the answers below:
You should read the following and learn a little bit about the advantages of a well designed innodb table and how best to use clustered indexes - only available with innodb !
Clustered indexes:
http://dev.mysql.com/doc/refman/5.0/en/innodb-index-types.html
http://www.xaprb.com/blog/2006/07/04/how-to-exploit-mysql-index-optimizations/
Previous answers of mine:
60 million entries, select entries from a certain month. How to optimize database?
MySQL and NoSQL: Help me to choose the right one
How to avoid "Using temporary" in many-to-many queries?
Optimal MySQL settings for queries that deliver large amounts of data?
Hope this helps :)
(EDITED)
A (year, week) or (week, year) index in big_table would definitely help this query.
I see you already have the second one, (week, year). I would prefer the first one but for this query it would not make much difference, if any.
I would add:
a composite index
(code_of_medium_table, year, week)in thebig_tableanda composite index
(code_of_small_table, indexed_field)in themedium_table.
Can you try it and post again the query plan?
Or (year, week, code_of_medium_table) and (indexed_field, code_of_small_table).
来源：https://stackoverflow.com/questions/6174246/mysql-innodb-query-performance