MySQL 8.0 information_schema系统库做了哪些改进?

程序员考拉

共 6588字,需浏览 14分钟

 ·

2020-07-11 08:49

2fb03e1411e9b694223a2ccb1323d24b.webp


目录


  • information_schema有何用?

  • mysql8.0 之前的查询方式

  • mysql8.0 开始的查询方式

  • 测试5.7和8.0不同版本访问I_S库的性能

  • 结论


information_schema有何用?


  • 通过I_S获取mysql的一些元数据信息

  • 获取表的数据文件、索引文件的大小、碎片情况、表行数、自增列增长情况等

  • 获取正在运行的事务有那些,是否有阻塞等

  • 获取当前mysql的连接processlist等等


mysql8.0 之前的查询方式


会在查询information_schema 某个表时创建临时表


  • 来自文件的元数据,扫描文件系统获取FRM文件的表定义

  • 存储引擎的详细信息,例如动态表统计信息

  • 来自MySQL服务器中全局数据结构的数据


在表的数量很多时,每次查询I_S会从文件系统中读取每个单独的FRM文件,使用更多的CPU周期来打开表并准备相关的内存数据结构


mysql8.0 开始的查询方式


  • 引入了基于InnoDB的本地数据字典表

  • 表中不在有FRM表定义文件

  • 所有数据库表的元数据都存储在事务数据字典表中

  • I_S中表被设计为数据字典表上的VIEW(有些还是临时表0_0)


消除了以下成本


查询INFORMATION_SCHEMA时创建的临时表
扫描文件系统目录以查找FRM文件


改进


利用MySQL优化器的全部功能,使用数据字典表上的索引来更好的查询


686b7b45ea55a3dc11657a25a4d8faf5.webp


mysql5.7中表文件


ll test*
Jul 10 10:52 testse.frm
Jul 10 10:52 testse.ibd


mysql8.0中表文件


ll test*
Jul 10 10:25 testse.ibd


mysql5.7 和mysql8.0 I_S 中tables 表的存在方式


mysql5.7.22 
show create table information_schema.tables\G
*************************** 1. row ***************************
       Table: TABLES
Create Table: CREATE TEMPORARY TABLE `TABLES` (
  `TABLE_CATALOG` varchar(512) NOT NULL DEFAULT '',
  `TABLE_SCHEMA` varchar(64) NOT NULL DEFAULT '',
  `TABLE_NAME` varchar(64) NOT NULL DEFAULT '',
  `TABLE_TYPE` varchar(64) NOT NULL DEFAULT '',
  `ENGINE` varchar(64) DEFAULT NULL,

mysql8.0.15
  show create table information_schema.tables\G
*************************** 1. row ***************************
                View: TABLES
         Create View: CREATE ALGORITHM=UNDEFINED DEFINER=`mysql.infoschema`@`localhost` SQL SECURITY DEFINER VIEW


  • mysql5.7 中I_S中tables表是以临时表的形式存在的(查询该表就会创建临时表,创建的临时表过多,可能会导致mysql占用的内存暴涨,出现OOM)

  • mysql8.0 中I_S中tables表以视图的形式存在(查询该视图,不会创建临时表,会使用到视图中表的索引)


mysql5.7中获取表大小情况


SELECT   table_name, CONCAT(FORMAT(data_length / 1024 / 1024, 2), 'M') AS dbdata_size, CONCAT(FORMAT(index_length / 1024 / 1024, 2), 'M') AS dbindex_size, CONCAT( FORMAT((data_length + index_length) / 1024 / 1024 / 1024,2), 'G') AS `db_size(G)`, AVG_ROW_LENGTH, table_rows, update_time FROM   information_schema.tables WHERE table_schema = 'test'   AND table_name = 'testse';
+------------+-------------+--------------+------------+----------------+------------+---------------------+
| table_name | dbdata_size | dbindex_size | db_size(G) | AVG_ROW_LENGTH | table_rows | update_time |
+------------+-------------+--------------+------------+----------------+------------+---------------------+
| testse | 0.02M | 0.02M | 0.00G | 862 | 19 | 2019-07-10 10:52:02 |
+------------+-------------+--------------+------------+----------------+------------+---------------------+


执行计划中出现了where,未用到索引(存储引擎检索数据后,server层进行过滤)
desc SELECT   table_name, CONCAT(FORMAT(data_length / 1024 / 1024, 2), 'M') AS dbdata_size, CONCAT(FORMAT(index_length / 1024 / 1024, 2), 'M') AS dbindex_size, CONCAT( FORMAT(data_length + index_length / 1024 / 1024 / 1024, 2   ), 'G') AS `db_size(G)`, AVG_ROW_LENGTH, table_rows, update_time FROM   information_schema.tables WHERE table_schema = 'test'   AND table_name = 'testse';
+----+-------------+--------+------------+------+---------------+-------------------------+---------+------+------+----------+---------------------------------------------------+
| id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra |
+----+-------------+--------+------------+------+---------------+-------------------------+---------+------+------+----------+---------------------------------------------------+
| 1 | SIMPLE | tables | NULL | ALL | NULL | TABLE_SCHEMA,TABLE_NAME | NULL | NULL | NULL | NULL | Using where; Open_full_table; Scanned 0 databases |
+----+-------------+--------+------------+------+---------------+-------------------------+---------+------+------+----------+---------------------------------------------------+


mysql8.0中获取表大小情况


SELECT table_name, CONCAT(FORMAT(data_length / 1024 / 1024, 2), 'M') AS dbdata_size, CONCAT(FORMAT(index_length / 1024 / 1024, 2), 'M') AS dbindex_size, CONCAT( FORMAT((data_length + index_length) / 1024 / 1024 / 1024,2), 'G') AS `db_size(G)`, AVG_ROW_LENGTH, table_rows, update_time FROM information_schema.tables WHERE table_schema = 'test'   AND table_name = 'testse';
+------------+-------------+--------------+------------+----------------+------------+---------------------+
| TABLE_NAME | dbdata_size | dbindex_size | db_size(G) | AVG_ROW_LENGTH | TABLE_ROWS | UPDATE_TIME |
+------------+-------------+--------------+------------+----------------+------------+---------------------+
| testse | 0.02M | 0.02M | 0.00G | 862 |         19 | 2019-07-10 10:25:16 |
+------------+-------------+--------------+------------+----------------+------------+---------------------+



能使用到数据字典表的索引
desc SELECT table_name, CONCAT(FORMAT(data_length / 1024 / 1024, 2), 'M') AS dbdata_size, CONCAT(FORMAT(index_length / 1024 / 1024, 2), 'M') AS dbindex_size, CONCAT( FORMAT((data_length + index_length) / 1024 / 1024 / 1024,2), 'G') AS `db_size(G)`, AVG_ROW_LENGTH, table_rows, update_time FROM information_schema.tables WHERE table_schema = 'test'   AND table_name = 'testse';
+----+-------------+-------+------------+--------+--------------------+------------+---------+-------------------------+------+----------+-------------+
| id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra |
+----+-------------+-------+------------+--------+--------------------+------------+---------+-------------------------+------+----------+-------------+
|
  1 | SIMPLE | cat | NULL | index | PRIMARY | name | 194 | NULL | 1 |   100.00 | Using index |
| 1 | SIMPLE | sch | NULL | eq_ref | PRIMARY,catalog_id | catalog_id | 202     | mysql.cat.id,const |    1 | 100.00 | Using index |
|
  1 | SIMPLE | tbl | NULL | eq_ref | schema_id | schema_id | 202 | mysql.sch.id,const | 1 |   100.00 | Using where |
| 1 | SIMPLE | stat | NULL | const | PRIMARY | PRIMARY | 388     | const,const |    1 | 100.00 | NULL |
|
  1 | SIMPLE | ts | NULL | eq_ref | PRIMARY | PRIMARY | 8 | mysql.tbl.tablespace_id | 1 |   100.00 | NULL |
| 1 | SIMPLE | col | NULL | eq_ref | PRIMARY | PRIMARY | 8       | mysql.tbl.collation_id |    1 | 100.00 | Using index |
+----+-------------+-------+------------+--------+--------------------+------------+---------+-------------------------+------+----------+-------------+


测试5.7和8.0不同版本访问I_S库的性能


机器


cat /etc/redhat-release | xargs echo '版本 ' && dmidecode -s system-product-name | xargs echo '是否虚拟化 ' && cat /proc/cpuinfo |grep "processor"|wc -l | xargs echo 'cpu核数 '
版本 CentOS Linux release 7.5.1804 (Core)
是否虚拟化 KVM
cpu核数 4


1、分别在mysql5.7和mysql8.0 中创建5个库,每个库中30个表(共1500个表),每个表记录数为10000


user=admin
passwd=admin
port=57222
host=127.0.0.1

#创建5个库,分别在每个库中创建30个表
for i in {1..5};do

mysql -u$user -p$passwd -P$port -h$host<  create database if not exists test_${i};

EOF

sysbench /usr/local/share/sysbench/oltp_read_write.lua --mysql_storage_engine=innodb --table-size=10000 --tables=30 --mysql-db=test_${i} --mysql-user=$user --mysql-password=$passwd --mysql-port=$port  --mysql-host=$host --threads=8 --time=10 --report-interval=1 --events=0 --db-driver=mysql prepare
done;


2、自定义访问I_S.tables表的sysbench脚本


cat tests/mytest.lua


require("oltp_common")


function thread_init(thread_id)
 drv=sysbench.sql.driver()
 con=drv:connect()
end


local function get_rand_db()
   return sysbench.rand.uniform(1, 5)
end



function event(thread_id)
local vid1
local dbprefix


vid1=get_rand_db()
dbprefix = "test_"

-- 生成5个db中的随机一个db,如 test_1 ,test_5
-- vid2=string.format("'%s%s'",test_,vid1)
vid2="'" .. dbprefix .. vid1 .. "'"

con:query("SELECT table_name, CONCAT(FORMAT(data_length / 1024 / 1024, 2), 'M') AS dbdata_size, CONCAT(FORMAT(index_length / 1024 / 1024, 2), 'M') AS dbindex_size, CONCAT( FORMAT((data_length + index_length) / 1024 / 1024 / 1024,2), 'G') AS `db_size(G)`, AVG_ROW_LENGTH, table_rows, update_time FROM information_schema.tables WHERE table_schema =" .. vid2)

end

function thread_done()
 con:disconnect()
end


3、脚本通过sysbench测试访问I_S.tables的qps


cat test_I_S_access.sh


require("oltp_common")


function thread_init(thread_id)
 drv=sysbench.sql.driver()
 con=drv:connect()
end


local function get_rand_db()
   return sysbench.rand.uniform(1, 5)
end



function event(thread_id)
local vid1
local dbprefix


vid1=get_rand_db()
dbprefix = "test_"

-- 生成5个db中的随机一个db,如 test_1 ,test_5
-- vid2=string.format("'%s%s'",test_,vid1)
vid2="'" .. dbprefix .. vid1 .. "'"

con:query("SELECT table_name, CONCAT(FORMAT(data_length / 1024 / 1024, 2), 'M') AS dbdata_size, CONCAT(FORMAT(index_length / 1024 / 1024, 2), 'M') AS dbindex_size, CONCAT( FORMAT((data_length + index_length) / 1024 / 1024 / 1024,2), 'G') AS `db_size(G)`, AVG_ROW_LENGTH, table_rows, update_time FROM information_schema.tables WHERE table_schema =" .. vid2)

end

function thread_done()
 con:disconnect()
end


4、mysql5.7和mysql8.0 主要配置


mysql5.7.22

5.7.22-log
innodb_buffer_pool_size 128M
innodb_log_buffer_size  64M
innodb_log_file_size    48M
binlog_format   ROW
innodb_flush_log_at_trx_commit  1
sync_binlog     1

mysql8.0.15

8.0.15
innodb_buffer_pool_size 128M
innodb_log_buffer_size  64M
innodb_log_file_size    48M
binlog_format   ROW
innodb_flush_log_at_trx_commit  1
sync_binlog     1


5、mysql5.7开启测试


bash test_I_S_acess.sh 57222  60


Shutting down MySQL.. SUCCESS!
Starting MySQL.. SUCCESS!
mysql version
5.7.22-log
Variable_name Value
Created_tmp_tables 1
sysbench 1.1.0 (using bundled LuaJIT 2.1.0-beta3)

Running the test with following options:
Number of threads: 40
Report intermediate results every 10 second(s)
Initializing random number generator from current time


Initializing worker threads...

Threads started!

[ 10s ] thds: 40 tps: 4765.73 qps: 4765.73 (r/w/o: 4765.73/0.00/0.00) lat (ms,95%): 18.95 err/s: 0.00 reconn/s:                                                                                                              0.00
[ 20s ] thds: 40 tps: 5409.00 qps: 5409.00 (r/w/o: 5409.00/0.00/0.00) lat (ms,95%): 17.95 err/s: 0.00 reconn/s:                                                                                                              0.00
[ 30s ] thds: 40 tps: 5154.45 qps: 5154.45 (r/w/o: 5154.45/0.00/0.00) lat (ms,95%): 18.61 err/s: 0.00 reconn/s:                                                                                                              0.00
[ 40s ] thds: 40 tps: 5383.50 qps: 5383.50 (r/w/o: 5383.50/0.00/0.00) lat (ms,95%): 17.63 err/s: 0.00 reconn/s:                                                                                                              0.00
[ 50s ] thds: 40 tps: 5456.11 qps: 5456.11 (r/w/o: 5456.11/0.00/0.00) lat (ms,95%): 17.63 err/s: 0.00 reconn/s:                                                                                                              0.00
[ 60s ] thds: 40 tps: 5458.66 qps: 5458.66 (r/w/o: 5458.66/0.00/0.00) lat (ms,95%): 17.63 err/s: 0.00 reconn/s:                                                                                                              0.00
SQL statistics:
    queries performed:
        read: 316322
        write: 0
        other: 0
        total:                           316322
    transactions:                        316322 (5270.99 per sec.)
    queries:                             316322 (5270.99 per sec.)
    ignored errors:                      0      (0.00 per sec.)
    reconnects:                          0      (0.00 per sec.)

Throughput:
    events/s (eps): 5270.9892
    time elapsed: 60.0119s
    total number of events:              316322

Latency (ms):
         min: 0.42
         avg:                                    7.59
         max: 858.18
         95th percentile: 17.95
         sum: 2399738.58

Threads fairness:
    events (avg/stddev): 7908.0500/53.41
    execution time (avg/stddev): 59.9935/0.00

mysql version
5.7.22-log
Variable_name Value
Created_tmp_tables 316327


cpu使用率接近100%


750a8179fed008a45130b22fcc7fcc58.webp


  • mysql5.7.22 40个并发线程 压力测试60秒,访问I_S.tables 平均qps为5k,并创建了31w个临时表!


6、mysql8.0开启测试


Shutting down MySQL.. SUCCESS!
Starting MySQL.... SUCCESS!
mysql version
8.0.15
Variable_name Value
Created_tmp_tables 1
sysbench 1.1.0 (using bundled LuaJIT 2.1.0-beta3)

Running the test with following options:
Number of threads: 40
Report intermediate results every 10 second(s)
Initializing random number generator from current time


Initializing worker threads...

Threads started!

[ 10s ] thds: 40 tps: 1283.76 qps: 1283.76 (r/w/o: 1283.76/0.00/0.00) lat (ms,95%): 41.10 err/s: 0.00 reconn/s: 0.00
[ 20s ] thds: 40 tps: 1286.43 qps: 1286.43 (r/w/o: 1286.43/0.00/0.00) lat (ms,95%): 39.65 err/s: 0.00 reconn/s: 0.00
[ 30s ] thds: 40 tps: 1333.00 qps: 1333.00 (r/w/o: 1333.00/0.00/0.00) lat (ms,95%): 37.56 err/s: 0.00 reconn/s: 0.00
[ 40s ] thds: 40 tps: 1280.79 qps: 1280.79 (r/w/o: 1280.79/0.00/0.00) lat (ms,95%): 40.37 err/s: 0.00 reconn/s: 0.00
[ 50s ] thds: 40 tps: 1313.57 qps: 1313.57 (r/w/o: 1313.57/0.00/0.00) lat (ms,95%): 38.94 err/s: 0.00 reconn/s: 0.00
[ 60s ] thds: 40 tps: 1222.95 qps: 1222.95 (r/w/o: 1222.95/0.00/0.00) lat (ms,95%): 42.61 err/s: 0.00 reconn/s: 0.00
SQL statistics:
    queries performed:
        read: 77248
        write: 0
        other: 0
        total:                           77248
    transactions:                        77248  (1286.95 per sec.)
    queries:                             77248  (1286.95 per sec.)
    ignored errors:                      0      (0.00 per sec.)
    reconnects:                          0      (0.00 per sec.)

Throughput:
    events/s (eps): 1286.9486
    time elapsed: 60.0242s
    total number of events:              77248

Latency (ms):
         min: 1.88
         avg:                                   31.08
         max: 692.98
         95th percentile: 40.37
         sum: 2400505.33

Threads fairness:
    events (avg/stddev): 1931.2000/51.17
    execution time (avg/stddev): 60.0126/0.00

mysql version
8.0.15
Variable_name Value
Created_tmp_tables 2


cpu使用率接近100%


234b4fc518142a0a054d4d1294f48db8.webp


  • mysql8.0.15 40个并发线程,压力测试60秒,访问I_S.tables 平均qps为1.2k,并创建了1个临时表!


结论


  • mysql8.0开始查询I_S中表不会再从文件系统(FRM, TRG,OPT文件)和myisam、innodb系统表读取元数据信息(8.0开始元数据信息统一存放在数据字典表中)

  • mysql8.0 访问I_S.tables不会创建临时表,这减少了内存暴涨的可能,但访问I_S.tables的qps大约是mysql5.7.22的1/5,访问速度没有mysql5.7.22的快

  • mysql8.0 访问I_S.tables 用户空间cpu消耗没有mysql5.7的高(没有创建临时表的缘故吧),但系统空间cpu消耗高于mysql5.7!



作者:jiaxin

出处:cnblogs.com/YangJiaXin/



浏览 22
点赞
评论
收藏
分享

手机扫一扫分享

分享
举报
评论
图片
表情
推荐
点赞
评论
收藏
分享

手机扫一扫分享

分享
举报