刚入职！就遇到上亿(MySQL)大表的优化

大家好，又见面了，我是全栈君。

背景

XX实例（一主一从）xxx告警中每天凌晨在报SLA报警，该报警的意思是存在一定的主从延迟（若在此时发生主从切换，需要长时间才可以完成切换，要追延迟来保证主从数据的一致性）

XX实例的慢查询数量最多（执行时间超过1s的sql会被记录），XX应用那方每天晚上在做删除一个月前数据的任务

分析

使用pt-query-digest工具分析最近一周的mysql-slow.log

pt-query-digest --since=148h mysql-slow.log | less

结果第一部分
刚入职！就遇到上亿(MySQL)大表的优化

最近一个星期内，总共记录的慢查询执行花费时间为25403s，最大的慢sql执行时间为266s，平均每个慢sql执行时间5s，平均扫描的行数为1766万

结果第二部分
刚入职！就遇到上亿(MySQL)大表的优化

select arrival_record操作记录的慢查询数量最多有4万多次，平均响应时间为4s，delete arrival_record记录了6次，平均响应时间258s。

select xxx_record语句

select arrival_record 慢查询语句都类似于如下所示，where语句中的参数字段是一样的，传入的参数值不一样
select count(*) from arrival_record where product_id=26 and receive_time between ‘2019-03-25 14:00:00’ and ‘2019-03-25 15:00:00’ and receive_spend_ms>=0\G

刚入职！就遇到上亿(MySQL)大表的优化
select arrival_record 语句在mysql中最多扫描的行数为5600万、平均扫描的行数为172万，推断由于扫描的行数多导致的执行时间长

查看执行计划

explain select count(*) from arrival_record where product_id=26 and receive_time between '2019-03-25 14:00:00' and '2019-03-25 15:00:00' and receive_spend_ms>=0\G;
*************************** 1. row ***************************
id: 1
select_type: SIMPLE
table: arrival_record
partitions: NULL
type: ref
possible_keys: IXFK_arrival_record
key: IXFK_arrival_record
key_len: 8
ref: const
rows: 32261320
filtered: 3.70
Extra: Using index condition; Using where
1 row in set, 1 warning (0.00 sec)

用到了索引IXFK_arrival_record，但预计扫描的行数很多有3000多w行

show index from arrival_record;
+----------------+------------+---------------------+--------------+--------------+-----------+-------------+----------+--------+------+------------+---------+---------------+
| Table | Non_unique | Key_name | Seq_in_index | Column_name | Collation | Cardinality | Sub_part | Packed | Null | Index_type | Comment | Index_comment |
+----------------+------------+---------------------+--------------+--------------+-----------+-------------+----------+--------+------+------------+---------+---------------+
| arrival_record | 0 | PRIMARY | 1 | id | A | 107990720 | NULL | NULL | | BTREE | | |
| arrival_record | 1 | IXFK_arrival_record | 1 | product_id | A | 1344 | NULL | NULL | | BTREE | | |
| arrival_record | 1 | IXFK_arrival_record | 2 | station_no | A | 22161 | NULL | NULL | YES | BTREE | | |
| arrival_record | 1 | IXFK_arrival_record | 3 | sequence | A | 77233384 | NULL | NULL | | BTREE | | |
| arrival_record | 1 | IXFK_arrival_record | 4 | receive_time | A | 65854652 | NULL | NULL | YES | BTREE | | |
| arrival_record | 1 | IXFK_arrival_record | 5 | arrival_time | A | 73861904 | NULL | NULL | YES | BTREE | | |
+----------------+------------+---------------------+--------------+--------------+-----------+-------------+----------+--------+------+------------+---------+---------------+
show create table arrival_record;
..........
arrival_spend_ms bigint(20) DEFAULT NULL,
total_spend_ms bigint(20) DEFAULT NULL,
PRIMARY KEY (id),
KEY IXFK_arrival_record (product_id,station_no,sequence,receive_time,arrival_time) USING BTREE,
CONSTRAINT FK_arrival_record_product FOREIGN KEY (product_id) REFERENCES product (id) ON DELETE NO ACTION ON UPDATE NO ACTION
) ENGINE=InnoDB AUTO_INCREMENT=614538979 DEFAULT CHARSET=utf8 COLLATE=utf8_bin |

该表总记录数约1亿多条，表上只有一个复合索引，product_id字段基数很小，选择性不好
传入的过滤条件 where product_id=26 and receive_time between ‘2019-03-25 14:00:00’ and ‘2019-03-25 15:00:00’ and receive_spend_ms>=0 没有station_nu字段，使用不到复合索引 IXFK_arrival_record的 product_id,station_no,sequence,receive_time 这几个字段
根据最左前缀原则，select arrival_record只用到了复合索引IXFK_arrival_record的第一个字段product_id，而该字段选择性很差，导致扫描的行数很多，执行时间长
receive_time字段的基数大，选择性好，可对该字段单独建立索引，select arrival_record sql就会使用到该索引

现在已经知道了在慢查询中记录的select arrival_record where语句传入的参数字段有 product_id，receive_time，receive_spend_ms，还想知道对该表的访问有没有通过其它字段来过滤了？

神器tcpdump出场的时候到了

使用tcpdump抓包一段时间对该表的select语句

tcpdump -i bond0 -s 0 -l -w - dst port 3316 | strings | grep select | egrep -i 'arrival_record' >/tmp/select_arri.log

获取select 语句中from 后面的where条件语句

IFS_OLD=$IFS
IFS=$'\n'
for i in `cat /tmp/select_arri.log `;do echo ${i#*'from'}; done | less
IFS=$IFS_OLD

arrival_record arrivalrec0_ where arrivalrec0_.sequence='2019-03-27 08:40' and arrivalrec0_.product_id=17 and arrivalrec0_.station_no='56742'
arrival_record arrivalrec0_ where arrivalrec0_.sequence='2019-03-27 08:40' and arrivalrec0_.product_id=22 and arrivalrec0_.station_no='S7100'
arrival_record arrivalrec0_ where arrivalrec0_.sequence='2019-03-27 08:40' and arrivalrec0_.product_id=24 and arrivalrec0_.station_no='V4631'
arrival_record arrivalrec0_ where arrivalrec0_.sequence='2019-03-27 08:40' and arrivalrec0_.product_id=22 and arrivalrec0_.station_no='S9466'
arrival_record arrivalrec0_ where arrivalrec0_.sequence='2019-03-27 08:40' and arrivalrec0_.product_id=24 and arrivalrec0_.station_no='V4205'
arrival_record arrivalrec0_ where arrivalrec0_.sequence='2019-03-27 08:40' and arrivalrec0_.product_id=24 and arrivalrec0_.station_no='V4105'
arrival_record arrivalrec0_ where arrivalrec0_.sequence='2019-03-27 08:40' and arrivalrec0_.product_id=24 and arrivalrec0_.station_no='V4506'
arrival_record arrivalrec0_ where arrivalrec0_.sequence='2019-03-27 08:40' and arrivalrec0_.product_id=24 and arrivalrec0_.station_no='V4617'
arrival_record arrivalrec0_ where arrivalrec0_.sequence='2019-03-27 08:40' and arrivalrec0_.product_id=22 and arrivalrec0_.station_no='S8356'
arrival_record arrivalrec0_ where arrivalrec0_.sequence='2019-03-27 08:40' and arrivalrec0_.product_id=22 and arrivalrec0_.station_no='S8356'
select 该表 where条件中有product_id,station_no,sequence字段，可以使用到复合索引IXFK_arrival_record的前三个字段

综上所示，优化方法为，删除复合索引IXFK_arrival_record，建立复合索引idx_sequence_station_no_product_id，并建立单独索引indx_receive_time

delete xxx_record语句

刚入职！就遇到上亿(MySQL)大表的优化

该delete操作平均扫描行数为1.1亿行，平均执行时间是262s

delete语句如下所示，每次记录的慢查询传入的参数值不一样

delete from arrival_record where receive_time < STR_TO_DATE('2019-02-23', '%Y-%m-%d')\G

执行计划

explain select * from arrival_record where receive_time < STR_TO_DATE('2019-02-23', '%Y-%m-%d')\G
*************************** 1. row ***************************
id: 1
select_type: SIMPLE
table: arrival_record
partitions: NULL
type: ALL
possible_keys: NULL
key: NULL
key_len: NULL
ref: NULL
rows: 109501508
filtered: 33.33
Extra: Using where
1 row in set, 1 warning (0.00 sec)

该delete语句没有使用索引（没有合适的索引可用），走的全表扫描，导致执行时间长

优化方法也是建立单独索引indx_receive_time(receive_time)

测试

拷贝arrival_record表到测试实例上进行删除重新索引操作
XX实例arrival_record表信息

du -sh /datas/mysql/data/3316/cq_new_cimiss/arrival_record*
12K /datas/mysql/data/3316/cq_new_cimiss/arrival_record.frm
48G /datas/mysql/data/3316/cq_new_cimiss/arrival_record.ibd
select count() from cq_new_cimiss.arrival_record;
+-----------+
| count() |
+-----------+
| 112294946 |
+-----------+
1亿多记录数

SELECT
table_name,
CONCAT(FORMAT(SUM(data_length) / 1024 / 1024,2),'M') AS dbdata_size,
CONCAT(FORMAT(SUM(index_length) / 1024 / 1024,2),'M') AS dbindex_size,
CONCAT(FORMAT(SUM(data_length + index_length) / 1024 / 1024 / 1024,2),'G') AS table_size(G),
AVG_ROW_LENGTH,table_rows,update_time
FROM
information_schema.tables
WHERE table_schema = 'cq_new_cimiss' and table_name='arrival_record';
+----------------+-------------+--------------+------------+----------------+------------+---------------------+
| table_name | dbdata_size | dbindex_size | table_size(G) | AVG_ROW_LENGTH | table_rows | update_time |
+----------------+-------------+--------------+------------+----------------+------------+---------------------+
| arrival_record | 18,268.02M | 13,868.05M | 31.38G | 175 | 109155053 | 2019-03-26 12:40:17 |
+----------------+-------------+--------------+------------+----------------+------------+---------------------+

磁盘占用空间48G，mysql中该表大小为31G，存在17G左右的碎片，大多由于删除操作造成的（记录被删除了，空间没有回收)

备份还原该表到新的实例中，删除原来的复合索引，重新添加索引进行测试

mydumper并行压缩备份

user=root
passwd=xxxx
socket=/datas/mysql/data/3316/mysqld.sock
db=cq_new_cimiss
table_name=arrival_record
backupdir=/datas/dump_$table_name
mkdir -p $backupdir 
  nohup echo `date +%T` && mydumper -u $user -p $passwd -S $socket -B $db -c -T $table_name -o $backupdir -t 32 -r 2000000 && echo `date +%T` &

并行压缩备份所花时间（52s）和占用空间（1.2G，实际该表占用磁盘空间为48G，mydumper并行压缩备份压缩比相当高！）

Started dump at: 2019-03-26 12:46:04
........

Finished dump at: 2019-03-26 12:46:56

du -sh   /datas/dump_arrival_record/
1.2G  /datas/dump_arrival_record/

拷贝dump数据到测试节点

scp -rp /datas/dump_arrival_record root@10.230.124.19:/datas

多线程导入数据

time myloader -u root -S /datas/mysql/data/3308/mysqld.sock -P 3308 -p root -B test -d /datas/dump_arrival_record -t 32

real 126m42.885s
user 1m4.543s
sys 0m4.267s

逻辑导入该表后磁盘占用空间

du -h -d 1 /datas/mysql/data/3308/test/arrival_record.*
12K /datas/mysql/data/3308/test/arrival_record.frm
30G /datas/mysql/data/3308/test/arrival_record.ibd
没有碎片，和mysql的该表的大小一致

cp -rp /datas/mysql/data/3308 /datas

分别使用online DDL和 pt-osc工具来做删除重建索引操作
先删除外键，不删除外键，无法删除复合索引，外键列属于复合索引中第一列

nohup bash /tmp/ddl_index.sh &
2019-04-04-10:41:39 begin stop mysqld_3308
2019-04-04-10:41:41 begin rm -rf datadir and cp -rp datadir_bak
2019-04-04-10:46:53 start mysqld_3308
2019-04-04-10:46:59 online ddl begin
2019-04-04-11:20:34 onlie ddl stop
2019-04-04-11:20:34 begin stop mysqld_3308
2019-04-04-11:20:36 begin rm -rf datadir and cp -rp datadir_bak
2019-04-04-11:22:48 start mysqld_3308
2019-04-04-11:22:53 pt-osc begin
2019-04-04-12:19:15 pt-osc stop
online ddl 花费时间为34 分钟，pt-osc花费时间为57 分钟，使用onlne ddl时间约为pt-osc工具时间的一半

做DDL 参考

刚入职！就遇到上亿(MySQL)大表的优化

实施

由于是一主一从实例，应用是连接的vip，删除重建索引采用online ddl来做。停止主从复制后，先在从实例上做（不记录binlog），主从切换，再在新切换的从实例上做（不记录binlog)

function red_echo () {

        local what="$*"
        echo -e "$(date +%F-%T)  ${what}"
}

function check_las_comm(){
    if [ "$1" != "0" ];then
        red_echo "$2"
        echo "exit 1"
        exit 1
    fi
}

red_echo "stop slave"
mysql -uroot -p$passwd --socket=/datas/mysql/data/${port}/mysqld.sock -e"stop slave"
check_las_comm "$?" "stop slave failed"

red_echo "online ddl begin"
 mysql -uroot -p$passwd --socket=/datas/mysql/data/${port}/mysqld.sock -e"set sql_log_bin=0;select now() as  ddl_start;ALTER TABLE $db_.\`${table_name}\` DROP FOREIGN KEY FK_arrival_record_product,drop index IXFK_arrival_record,add index idx_product_id_sequence_station_no(product_id,sequence,station_no),add index idx_receive_time(receive_time);select now() as ddl_stop" >>${log_file} 2>& 1
 red_echo "onlie ddl stop"
 red_echo "add foreign key"
 mysql -uroot -p$passwd --socket=/datas/mysql/data/${port}/mysqld.sock -e"set sql_log_bin=0;ALTER TABLE $db_.${table_name} ADD CONSTRAINT _FK_${table_name}_product FOREIGN KEY (product_id) REFERENCES cq_new_cimiss.product (id) ON DELETE NO ACTION ON UPDATE NO ACTION;" >>${log_file} 2>& 1
 check_las_comm "$?" "add foreign key error"
 red_echo "add foreign key stop"

red_echo "start slave"
mysql -uroot -p$passwd --socket=/datas/mysql/data/${port}/mysqld.sock -e"start slave"
check_las_comm "$?" "start slave failed"

执行时间

2019-04-08-11:17:36 stop slave
mysql: [Warning] Using a password on the command line interface can be insecure.
ddl_start
2019-04-08 11:17:36
ddl_stop
2019-04-08 11:45:13
2019-04-08-11:45:13 onlie ddl stop
2019-04-08-11:45:13 add foreign key
mysql: [Warning] Using a password on the command line interface can be insecure.
2019-04-08-12:33:48 add foreign key stop
2019-04-08-12:33:48 start slave

再次查看delete 和select语句的执行计划

explain select count(*) from arrival_record where receive_time < STR_TO_DATE('2019-03-10', '%Y-%m-%d')\G
*************************** 1. row ***************************
id: 1
select_type: SIMPLE
table: arrival_record
partitions: NULL
type: range
possible_keys: idx_receive_time
key: idx_receive_time
key_len: 6
ref: NULL
rows: 7540948
filtered: 100.00
Extra: Using where; Using index
explain select count(*) from arrival_record where product_id=26 and receive_time between '2019-03-25 14:00:00' and '2019-03-25 15:00:00' and receive_spend_ms>=0\G;
*************************** 1. row ***************************
id: 1
select_type: SIMPLE
table: arrival_record
partitions: NULL
type: range
possible_keys: idx_product_id_sequence_station_no,idx_receive_time
key: idx_receive_time
key_len: 6
ref: NULL
rows: 291448
filtered: 16.66
Extra: Using index condition; Using where
都使用到了idx_receive_time 索引，扫描的行数大大降低

索引优化后

delete 还是花费了77s时间

delete from arrival_record where receive_time < STR_TO_DATE('2019-03-10', '%Y-%m-%d')\G

刚入职！就遇到上亿(MySQL)大表的优化

delete 语句通过receive_time的索引删除300多万的记录花费77s时间*

delete大表优化为小批量删除

应用端已优化成每次删除10分钟的数据（每次执行时间1s左右），xxx中没在出现SLA（主从延迟告警）
刚入职！就遇到上亿(MySQL)大表的优化

另一个方法是通过主键的顺序每次删除20000条记录

#得到满足时间条件的最大主键ID
#通过按照主键的顺序去 顺序扫描小批量删除数据
#先执行一次以下语句
 SELECT MAX(id) INTO @need_delete_max_id FROM `arrival_record` WHERE receive_time<'2019-03-01' ;
 DELETE FROM arrival_record WHERE id<@need_delete_max_id LIMIT 20000;
 select ROW_COUNT();  #返回20000


#执行小批量delete后会返回row_count(), 删除的行数
#程序判断返回的row_count()是否为0，不为0执行以下循环，为0退出循环，删除操作完成
 DELETE FROM arrival_record WHERE id<@need_delete_max_id LIMIT 20000;
 select ROW_COUNT();
#程序睡眠0.5s