MySQL Webinar 2/4 Performance tuning, hardware, optimisation

Installation, Configuration
and performance tuning
Mark Swarbrick
Principle Presales Consultant
MySQL Webinar Series 2/4

Program
Agenda
§  Hardware Specification
§  Setting up Replication
§  Performance tuning

The following is intended to outline our general product direction. It is
intended for information purposes only, and may not be incorporated
into any contract. It is not a commitment to deliver any material, code,
or functionality, and should not be relied upon in making purchasing
decision. The development, release, and timing of any features or
functionality described for Oracle’s products remains at the sole
discretion of Oracle.
Safe Harbor Statement

§  Hardware Specification

MySQL Hardware / Software Platforms
MySQL is supported on a variety of hardware /
software platforms:
Oracle Linux
Oracle Solaris
Red Hat
Canonical (Ubuntu)
SUSE
Debian
Microsoft Windows XP/Vista/7/2003/2008/2012/server
FreeBSD
HPUX
Other Linux – Fedora / OpenSUSE / CentOS etc

MySQL is supported on a variety of hardware /
software platforms:
X86
X86_64
X86_IA64
SPARC_32
SPARC_64
IBM Power5_64
PA-RISC 2.0
We will support with best efforts on other hardware platforms such
as ARM. For example MySQL Cluster compiled to run on a
raspberry pi!

Copyright Oracle Corporation 2014 | <date>8
Scaling on dense,
multi-core, multi-thread
servers
•  10s - 100GBs RAM
•  SSDs
Scale across cores
within a single instance
You can get a long
way with MySQL 5.6!
The Path to Scalability
Scaling-Up can take you a long way

Server:
Having the fastest CPU is still beneficial. Can use as many as 32 cores!
OS choice and tuning is important, but Linux / Solaris or windows all give good
performance.
Storage:
Don’t use slow disks
Flash memory (SSD etc) helps for random access (reads are the most costly)
FS is important – ZFS, UFS, QFS, VxFS, EXT3, EXT4 XFS etc..
Use of O_Direct if O/S supports it
Don’t forget the network! Faster is better!

Hardware: The Perfect MySQL Server
•  The more cores the better (especially for 5.5 and later)
•  x86_64 - 64 bit for more memory is important
–  The more the better
•  Fast HD (10-15k RPM SATA) or NAS/SAN……
–  RAID 10 for most, RAID 5 OK if very read intensive
–  Hardware RAID battery backed up cache critical!
–  More disks are always better! - 4+ recommended, 8-16 can increase IO
•  …Or SSD (for higher throughput)
–  Intel, Fusion-IO good choices; good option for Slaves
•  At least 2 x NICs for redundancy
•  Slaves should be as powerful as the Master

MySQL HA Solutions
Note that MySQL Fabric is currently at Release Candidate Status

Copyright © 2013, Oracle and/or its affiliates. All rights reserved.13
Why Replicate?
§  Duplicates database from a “master” to a “slave”
§  Redundant copies of the data provide foundation for High Availability
§  Scale out by distributing queries across the replication cluster
Master
Slaves
Web / App Servers
Writes & Reads Reads

Slave
Database
§  Session thread: processes queries from
the application – writes data to master
database & associated events to binary
log
§  Dump thread: reads events from binary
log and sends them to a slave
§  I/O thread: receives replication events
and stores them in slave’s relay log
§  SQL thread: reads replication events from
slave’s relay log and applies them to
slave database
Session
Binary
Log
Master
Database
Dump I/O
Relay
Log
SQL
Replication Workflow

ReplicationReplication Solutions

Async Replication

Semi-sync Replication

MySQL 5.6: Transforming Replication
•  Multi-Threaded Slaves
•  Binary Log Group Commit
•  Optimized Row-Based Replication
PERFORMANCE
•  Global Transaction Identifiers
•  Replication Failover & Admin Utilities
•  Crash Safe Slaves & Binlogs
FAILOVER & RECOVERY
•  Replication Event Checksums
DATA INTEGRITY
•  Replication Utilities
•  Time-Delayed Replication
•  Remote Binlog Backup
•  Informational Log Events
•  Server UUIDs
DEV/OPS AGILITY

5x Higher Performance with MySQL 5.6
0 5 10
QPS 58.11 144.4 282.53
0
50
100
150
200
250
300
QueriesperSecond
Worker Threads
Multi-Threaded Slave Performance

Master
GTID=123456
GTID=123456
GTID=123456 GTID=123456
Global Transaction Identifiers
§  Simple to track & compare replication across the farm
§  Unique identifier for each transaction written to the Binlog
§  Automatically identify the most up-to-date slave for failover
§  Deploy n-tier replication hierarchies

Time Delayed Replication - configuring
slave2> CHANGE MASTER TO
-> MASTER_HOST = 'localhost‘,
-> MASTER_PORT = 3306,
-> MASTER_USER = 'repl_user‘,
-> MASTER_PASSWORD = ‘sun123!‘,
-> MASTER_DELAY = 600;
slave2> START SLAVE;

MySQL Utilities
§  Automate common Dev/Ops tasks
§  Replication: provisioning, testing, monitoring and failover
§  Database comparisons: consistency checking
§  Database administration: users, connections, tables, etc
§  New utilities in development, i.e. log analysis
§  Implemented as Python scripts, plug-in for MySQL Workbench
§  Extensible to include custom scripting
§  Resources: Documentation & Community Forum
§  http://dev.mysql.com/doc/workbench/en/mysql-utils-man.html
§  http://forums.mysql.com/list.php?155

Utility Workflow for Replication
§  Replicate: Initiates Replication to the new slave
§  Check: Validates the replication configuration and operation
§  Show: Display Replication topology
§  Fail-Over & Admin: Detects and failovers (or switches) from master to
slave. Status monitoring
Check Show
Fail-Over
& AdminReplicate

mysqlreplicate Utility
§  Starts replication process
§  User can provide login and
connection parameters for the master
§  Configurable replication start points:
current position / GTID, specific
event, entire binlog
§  Checks storage engine compatibility
§  Enables fast, simple scaling
New Slave
Login,
Connection &
Start-Point
Master
Replication configured,
binlog position
Repl

mysqlreplicate example
$ mysqlreplicate --master=root:billy@localhost:3306
--slave=root:billy@localhost:3307
# master on localhost: ... connected.
# slave on localhost: ... connected.
Repl

mysqlrplshow Utility
§  Discovers & displays replication
topology
§  Shows slaves attached to each
master
§  Labels each slave with hostname
and port number
§  Displays multiple topologies,
including hierarchical and circular
§  Displays results in a graph or list
§  Enables user to monitor topology
on-demand
$ mysqlrplshow --master=root@localhost:3311
# master on localhost: ... connected.
# Finding slaves for master: localhost:3311
# Replication Topology Graph::
localhost:3306 (MASTER)
|
+--- localhost:3307 - (SLAVE)
|
+--- localhost:3308 - (SLAVE + MASTER)
|
+--- localhost:3313 - (SLAVE)
Show

mysqlrplcheck Utility
§  Tests replication on master and slave to ensure proper configuration of
replication
§  Checks binlog is enabled and displays any configured exceptions
§  Checks slave access and privileges to master
§  Checks slave connection status
§  Checks for conflicts in master.info file (or mysql.slave_master_info table)
§  Checks storage engine compatibility
§  Tests executed sequentially
§  Reports status of each test
§  Simple verification of deployment & fast fault resolution
Check

mysqlrplcheck example
$ mysqlrplcheck --master=root@host1:3310
--slave=root@host2:3311
# master on host1: ... connected.
# slave on host2: ... connected. Test Description Status
-------------------------------------------------------------
Checking for binary logging on master [pass]
Are there binlog exceptions? [pass]
Replication user exists? [pass]
Checking server_id values [pass]
Is slave connected to master? [pass]
Check master information file [pass]
Checking InnoDB compatibility [pass]
Checking storage engines compatibility [pass]
Checking lower_case_table_names settings [pass]
Checking slave delay (seconds behind master) [pass]
# ...done.
Check

Replication Administration Utility
§  Perform switchover to eliminate
downtime during planned maintenance
§  Start and stop slaves
§  Slave discovery & monitoring
§  Slave status, thread status
§  Replication processing, including any lag
§  Configure slave promotion policies
Master
Slaves
Administration
Utility
Status &
Switchover
Admin

Planned switchover to new master with
mysqlrpladmin
$ mysqlrpladmin --master=root@rod:3306
--slaves=root@jane:3306,root@jane:3307,root@freddy:3306
--new-master=root@jane:3306 --demote-master switchover
# Performing switchover from master at rod:3306 to slave at jane:3306.
# Checking candidate slave prerequisites.
# Waiting for slaves to catch up to old master.
# Stopping slaves.
# Performing STOP on all slaves.
# Demoting old master to be a slave to the new master.
# Switching slaves to new master.
# Starting all slaves.
# Performing START on all slaves.
# Checking slaves for errors.
# Switchover complete.
# ...done.
Switch-
Over

§  Enabling self-healing replication topologies
§  Automated failover & recovery
§  mysqlfailover Utility
§  Switchover & administration
§  mysqlrpladmin Utility
§  Fully configurable slave promotion policies
§  Delivers HA within the core MySQL
distribution
HA UtilitiesMonitoring
Failed
Master
Slaves
Promoted
Master
Fail-Over
& Admin
Replication HA Utilities

mysqlfailover – monitoring resumes
MySQL Replication Monitor and Failover Utility
Failover Mode = auto Next Interval = Mon Mar 19 16:05:12 2012
Master Information
------------------
Binary Log File Position Binlog_Do_DB Binlog_Ignore_DB
mysql-bin.000001 1117
GTID Executed Set
A0F7E82D-3554-11E2-9950-080027685B56:1-5
UUIDs
+------------+-------+---------+--------+------------+---------+
| host | port | role | state | gtid_mode | health |
+------------+-------+---------+--------+------------+---------+
| jane | 3306 | MASTER | UP | ON | OK |
| jane | 3307 | SLAVE | UP | ON | OK |
| freddy | 3306 | SLAVE | UP | ON | OK |
+------------+-------+---------+--------+------------+---------+
Fail-Over

mysqlfailover – master failed
Failover starting...
# Candidate slave jane:3306 will become the new master.
# Preparing candidate for failover.
# Creating replication user if it does not exist.
# Stopping slaves.
# Performing STOP on all slaves.
# Switching slaves to new master.
# Starting slaves.
# Performing START on all slaves.
# Checking slaves for errors.
# Failover complete.
# Discovering slaves for master at jane:3306
Failover console will restart in 5 seconds.
Fail-Over

An extensible and easy-to-
use framework for
managing a farm of MySQL
server supporting high-
availability and sharding
MySQL Fabric

MySQL Fabric Framework (HA)
All Data
Primary Secondary
Extra Read
Replicas
MySQL Fabric
Controller
SQL Queries
State &
Routing Info
HA Group
Coordination
and Control

MySQL Replication & MySQL Fabric HA
§  MySQL Replication is the initial implementation used in HA Groups
–  PRIMARY = Replication Master & receives all writes
§  Failover
–  MySQL Fabric detects failure of PRIMARY/Master
–  Selects a SECONDARY/Slave and promotes it
–  Updates State Store
–  Updated state fetched by Fabric-aware connectors
& how this effects failover

High-Availability Group Concept
§  Abstract Concept
–  Set of servers
–  Server attributes
§  Connector Attributes
–  Connection information
–  Mode: read-only, read-write, ...
–  Weight: distribute load
§  Management Attributes
–  State: state/role of the server State: Primary
Mode: Read-Write
Host: server-1.example.com

Create HA Groups and add Servers
§  Define a group
mysqlfabric group create my_group
§  Add servers to group
mysqlfabric group add my_group server1.example.com
mysqlfabric group add my_group server2.example.com

Create HA Groups and add Servers
§  Promote one server to be primary
mysqlfabric group promote my_group
§  Tell failure detector to monitor group
mysqlfabric group activate my_group

§  Performance Tuning

The World’s Most Popular Open Source DatabaseCopyright 2010 Oracle
Rules of Tuning 

•  Never make a change in production first
•  Have a good benchmark or reliable load
•  Start with a good baseline
•  Only change 1 thing at a time
–  identify a set of possible changes
–  try each change separately
–  try in combinations of 2, then 3, etc.
•  Monitor the results
–  Query performance - query analyzer, slow query log, etc.
•  throughput
•  single query time
•  average query time
–  CPU - top, vmstat, dstat
–  IO - iostat, top, vmstat, bonnie++, dstat
–  Network bandwidth
•  Document and save the results

Were do I find a benchmark?
•  Make your own
–  Can use general query log output
–  Could use MySQL Proxy and TCP Dump
•  DBT2
–  http://osdldbt.sourceforge.net/
–  http://samurai-mysql.blogspot.com/2009/03/settingup-dbt-2.html
•  mysqlslap MySQL 5.1 +
–  http://dev.mysql.com/doc/refman/5.1/en/mysqlslap.html
•  SysBench
–  http://sysbench.sourceforge.net/
•  supersmack
–  http://vegan.net/tony/supersmack/
•  mybench
–  http://jeremy.zawodny.com/mysql/mybench/

InnoDB Tuning 

•  Unlike MyISAM - InnoDB uses a single cache for both index and data
–  innodb_buffer_pool_size - should be 70-80% of available
memory.
–  It is not uncommon for this to be very large, i.e. 34GB on a system with
40GB of memory
–  Make sure its not set so large as to cause swapping!
–  mysql>show status like 'Innodb_buffer%' ;
•  InnoDB can use direct IO on systems that support it - Linux, FreeBSD,
and Solaris
–  innodb_flush_method = O_DIRECT
•  For more InnoDB tuning see
–  http://dev.mysql.com/doc/refman/5.1/en/innodb-tuning-
troubleshooting.html

DBT-2 (W200) Transactions per Minute %user %iowait
Buffer pool 1G 1125.44 2% 30%
Buffer pool 2G 1863.19 3% 28%
Buffer pool 5G 4385.18 5.5% 33%
Buffer pool 30G
(All data in cache)
36784.76 36% 8%
Cache hot application data in memory 

•  DBT-2 benchmark (write intensive)
•  20-25GB hot data (200 warehouses, running 1 hour)
•  Nehalem 2.93GHz x 8 cores, MySQL 5.5.2, 4 RAID1+0 HDDs
•  RAM size affects everything. Not only for SELECT, but also for INSERT/UPDATE/DELETE
–  INSERT: Random reads/writes happen when inserting into indexes in random order
–  UPDATE/DELETE: Random reads/writes happen when modifying records

Connections 

•  MySQL Caches the threads used by a connection
–  thread_cache_size - Number of threads to cache
–  Setting this to 100 or higher is not unusual
•  Monitor Threads_created to see if this is an issue
–  Counts connections not using the thread cache
–  Should be less than 1-2 a minute
–  Usually only an issue if more than 1-2 a second
•  Only an issue if you create and drop a lot of connections, i.e. PHP
•  Overhead is usually about 250k per thread
•  Aborted_clients -
http://dev.mysql.com/doc/refman/5.1/en/communication-errors.html
•  Aborted_connections -
http://dev.mysql.com/doc/refman/5.1/en/communication-errors.html

Sessions 

•  Some session variables control space allocated by each session (per connection)
–  Setting these to small can give bad performance
–  Setting these too large can cause the server to swap!
–  Can be set by connection
•  SET SORT_BUFFER_SIZE=1024*1024*128
–  Set small by default, increase in connections that need it
•  sort_buffer_size - Used for ORDER BY, GROUP
• 
BY, SELECT DISTINCT, UNION DISTINCT
–  Monitor Sort_merge_passes < 1-2 an hour optimal
–  Usually a problem in a reporting or data warehouse database
•  Other important session variables
–  read_rnd_buffer_size - Set to 1/2 sort_buffer_size
–  join_buffer_size - Watch Select_full_join
–  read_buffer_size - Used for full table scans, watch Select_scan
–  tmp_table_size - Max temp table size in memory, watch Created_tmp_disk_tables

Query Cache 

•  MySQL’s ‘Jekyll and Hyde’ of performance tuning options, when it is useful it really
helps, when it hurts, it really hurts
•  MySQL Query Cache caches both the query and the full result set
–  query_cache_type - Controls behaviour
•  0 or OFF - Not used (buffer may still be allocated)
•  1 or ON cache all unless SELECT SQL_NO_CACHE (DEFAULT)
•  2 or DEMAND cache none unless SELECT SQL_CACHE
–  query_cache_size - Determines the size of the cache
•  mysql> show status like 'Qc%' ;
•  Gives great performance if:
–  Identical queries returning identical data are used often
–  No or rare inserts, updates or deletes
•  Best Practice
–  Set to DEMAND
–  Add SQL_CACHE to appropriate queries

Indexes in MySQL 

•  Indexes allow for faster access to data
•  Data accessed via an index is usually in sorted order
•  Unique or Primary - Must refer to only one record
•  Non-Unique - May refer to many records
•  Can be on one or more columns
–  CREATE INDEX IDX ON TAB1(col1,col2,col3) ;
•  Can use prefix index for
–  CHAR, VARCHAR, BINARY, and VARBINARY
–  CREATE INDEX PRE ON TAB1(COL1(10)) ;
–  Prefix is in bytes, not characters
•  Very useful for large strings
•  Works best when leading part of column is selective

Index Best Practices 

•  Too many indexes can slow down inserts/deletes
–  Use only the indexes you must have
–  Check often
–  mysql>show create table tabname ;
•  Don’t duplicate leading parts of compound keys
–  index key123 (col1,col2,col3)
–  index key12 (col1,col2) <- Not needed!
–  index key1 (col1) <-- Not needed!
•  Use prefix indexes on large keys
•  Best indexes are 16 bytes/chars or less (> 16 text?)
•  Indexes bigger than 32 bytes/chars should be looked at very closely
–  should have there own cache if in MyISAM
•  For large strings that need to be indexed, i.e. URLs, consider using a separate
column using the MySQL MD5 to create a hash key and index on it instead

Explain - Details 

•  Tables are accessed from top to bottom
•  Columns
–  Select Type - SELECT if no Union or Subquery
–  Table, uses aliases
–  Type - Most common ref or eq_ref
–  Possible Keys - Indexes the optimizer is considering
–  Key = The index the optimizer chose
–  Ref - What column in what table (using alias) is referenced by the index
–  Rows - Estimated number of rows per reference
•  Multiple these to get overall cost
•  There are more values, see:
•  http://dev.mysql.com/doc/refman/5.6/en/using-explain.html

mysql> explain select C.Name, Y.Name, Y.Population, Language from Country as C, City as Y, CountryLanguage as L where Y.Name = C.Name and
L.CountryCode = Y.CountryCode and C.Name = 'Macao' ;
+----+-------------+-------+------+---------------+---------+---------+---------------------+------+--------------------------+
| id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra
|+----+-------------+-------+------+---------------+---------+---------+---------------------+------+--------------------------+
| 1 | SIMPLE | C | ref | c2 | c2 | 52 | const | 1 | Using where; Using index |
| 1 | SIMPLE | Y | ref | c2 | c2 | 35 | const | 1 | Using where |
| 1 | SIMPLE | L | ref | PRIMARY | PRIMARY | 3 | world.Y.CountryCode | 9 | Using index
|+----+-------------+-------+------+---------------+---------+---------+---------------------+------+--------------------------+
3 rows in set (0.00 sec)
More Explain 

•  alter table Country add index c2 (Name) ;
•  alter table City add index c2 (Name) ;
•  The original cost was 239 * 4079 * 9 = 8,773,929
•  The new cost is 1 * 1 * 9 = 9

Queries 

•  Often the # 1 issue in overall performance
•  Always, Always have your slow query log on!
–  http://dev.mysql.com/doc/refman/5.1/en/slow-query-log.html
–  Use: log_queries_not_using_indexes
–  Check it regularly
–  Use mysqldumpslow : http://dev.mysql.com/doc/refman/5.1/en/mysqldumpslow.html
–  Best practice is to automate running mysqldumpslow every morning and email results
to DBA, DBDev, etc.
•  Understand and use EXPLAIN
–  http://dev.mysql.com/doc/refman/5.1/en/using-explain.html
•  Select_scan - Number of full table scans
•  Select_full_join - Joins without indexes
•  MySQL Query Analyzer
–  http://www.mysql.com/products/enterprise/query.html

Schemas 

•  Size = performance, smaller is better
–  Size right! Do not automatically use 255 for VARCHAR
•  Temp tables, most caches, expand to full size
•  Use “procedure analyse” to determine the optimal types given the values in your
table
–  http://dev.mysql.com/doc/refman/5.1/en/procedure-analyse.html
–  mysql> select * from tab procedure analyse (64,2000) G
•  Consider the types:
–  enum : http://dev.mysql.com/doc/refman/5.1/en/enum.html
–  set : http://dev.mysql.com/doc/refman/5.1/en/set.html
•  Compress large strings
–  Use the MySQL COMPRESS and UNCOMPRESS functions
–  Very important in InnoDB 5.5, now with barracuda not so bad

Partitioning can help performance
l  Single inserts and selects faster
l  Range selects faster
l  More efficiently split the data across different paths
l  Historical data stored efficiently
l  delete large chunks of data (drop partition)
l  Move from table to partition easy 5.6

MySQL 5.6 Performance Schema
•  PERFORMANCE_SCHEMA presents low level
MySQL performance information
•  Data can be cleared
•  Filters with WHERE are allowed
•  Must be enabled with --performance_schema
mysql> SELECT EVENT_ID, EVENT_NAME, TIMER_WAIT
-> FROM EVENTS_WAITS_HISTORY WHERE THREAD_ID = 13
-> ORDER BY EVENT_ID;
+----------+-----------------------------------------+------------+
| EVENT_ID | EVENT_NAME | TIMER_WAIT |
+----------+-----------------------------------------+------------+
| 86 | wait/synch/mutex/mysys/THR_LOCK::mutex | 686322 |
| 87 | wait/synch/mutex/mysys/THR_LOCK_malloc | 320535 |
| 90 | wait/synch/mutex/sql/LOCK_plugin | 614673 |
| 91 | wait/synch/mutex/sql/LOCK_open | 659925 |
| 92 | wait/synch/mutex/sql/THD::LOCK_thd_data | 494001 |
| 95 | wait/synch/mutex/mysys/LOCK_alarm | 312993 |
+----------+-----------------------------------------+------------+
mysql> UPDATE SETUP_INSTRUMENTS
-> SET ENABLED = 'NO'
-> WHERE NAME = 'wait/synch/mutex/myisammrg/MYRG_INFO::mutex';
mysql> UPDATE SETUP_CONSUMERS
-> SET ENABLED = 'NO' WHERE NAME = 'file_summary_by_instance';
© 2010 Oracle Corporation – Proprietary and Confidential

•  Statements/Stages
•  What are my most resource intensive queries? Where do they spend time?
•  Table/Index I/O, Table Locks
•  Which application tables/indexes cause the most load or contention?
•  Network I/O
•  What is the network load like? How long do sessions idle?
•  Users/Hosts/Accounts
•  Which application users, hosts, accounts are consuming the most resources?
•  Summaries
•  Aggregated statistics grouped by thread, user, host, account or object
MySQL 5.6.3 – Performance Schema Instrumentation
Improved Database Profiling/Application Tuning

Next Steps
Learn More
Evaluate MySQL 5.6 Book a Workshop
HA Solutions
Guide
http://bit.ly/NB8up1
Download Today
http://dev.mysql.com/
downloads/mysql/
High Availability
Jumpstart
http://www.mysql.com/
consulting/
Copyright 2011 Oracle Corporation

mark.swarbrick@oracle.com

MySQL Webinar 2/4 Performance tuning, hardware, optimisation

Recommended

More Related Content

What's hot (20)

Viewers also liked (19)

Similar to MySQL Webinar 2/4 Performance tuning, hardware, optimisation (20)

More from Mark Swarbrick (20)

Recently uploaded (20)

MySQL Webinar 2/4 Performance tuning, hardware, optimisation