Parallel replication of MariaDB 10--Delay test results

Test parameters:

Sysbench--test=/root/sysbench0.5/sysbench/tests/db/insert.lua--mysql-table-engine=innodb--oltp-table-size= 1000000--max-requests=0--max-time=300--num-threads=16--oltp-tables-count=10--report-interval=10--mysql-host= 10.8.8.100--mysql-port=3312--mysql-user=admin--mysql-password=123456--mysql-db=test Run

10 1 million-row table, concurrent 16 threads, pure insert operation.

MySQL parameters:

Sync_binlog = 0innodb_flush_log_at_trx_commit = 0innodb_flush_method = O_directinnodb_flush_neighbors = 1innodb_buffer _pool_size = 10ginnodb_io_capacity = 500innodb_log_file_size = 512minnodb_log_files_in_group = 3innodb_log_buffer_size = 16M

Normal replication delay:

              Master_Log_File:  mysql-bin.000004          read_master_log_pos:  567748597               relay_log_ file: host-100-relay-bin.000005                 relay_log_pos: 66853042        relay_ master_log_file: mysql-bin.000003              slave_io_running: yes            slave_ sql_running: yes               replicate_do_db:           replicate_ignore_db:             replicate_do_table:        replicate_ Ignore_table:       replicate_wild_do_table:   replicate_ wild_ignore_table:                     Last_Errno: 0                    Last_Error:                   Skip_Counter: 0           exec_master_log_pos: 469504741

There's 1117MB left, not after.

MariaDB [(None)]> Select 1024-469504741/1024/1024+567748597/1024/1024;+--------------------------------------- -------+|                                1024-469504741/1024/1024+567748597/1024/1024 |+----------------------------------------------+| 1117.69264221 |+----------------------------------------------+1 row in Set (0.00 sec)

Parallel replication (8 concurrent threads, Gtid mode not turned on) delay:

              Master_Log_File:  mysql-bin.000022          read_master_log_pos:  333773843               relay_log_ file: host-100-relay-bin.000073                 relay_log_pos: 134217580        relay_ master_log_file: mysql-bin.000021              slave_io_running: yes            slave_ sql_running: yes               replicate_do_db:           replicate_ignore_db:              replicate_do_table:        replicate_ Ignore_table:       replicate_wild_do_table:   replicate_ wild_ignore_table:                     Last_Errno: 0                    Last_Error:                   Skip_Counter: 0           exec_master_log_pos: 722396020

There's 653MB left, not after.

MariaDB [(None)]> Select 1024-722396020/1024/1024+333773843/1024/1024;+--------------------------------------- -------+|                                 1024-722396020/1024/1024+333773843/1024/1024 |+----------------------------------------------+| 653.38101101 |+----------------------------------------------+1 row in Set (0.00 sec)

Parallel replication (8 concurrent threads, turn on Gtid mode) delay:

              Master_Log_File:  mysql-bin.000002          read_master_log_pos:  365095829               relay_log_ file: host-100-relay-bin.000008                 relay_log_pos: 134217572        relay_ master_log_file: mysql-bin.000001              slave_io_running: yes            slave_ sql_running: yes               replicate_do_db:           replicate_ignore_db:              replicate_do_table:        replicate_ Ignore_table:       replicate_wild_do_table:   replicate_ wild_ignore_table:                     Last_Errno: 0                    Last_Error:                   Skip_Counter: 0           exec_master_log_pos: 811005100

There's 598MB left, not after.

MariaDB [(None)]> Select 1024-811005100/1024/1024+365095829/1024/1024;+--------------------------------------- -------+|                                 1024-811005100/1024/1024+365095829/1024/1024 |+----------------------------------------------+| 598.74778080 |+----------------------------------------------+1 row in Set (0.00 sec)

Conclusion:

In my test, parallel replication is more than twice times faster than normal replication.

This article is from the "Spring Yang Technical column" blog, please be sure to keep this source http://hcymysql.blog.51cto.com/5223301/1433108