MySQL 5.6, news in 5.7 and our HA options

Copyright © 2014, Oracle and/or its affiliates. All rights reserved. |
Introduction to MySQL
Ted Wennmark, consultant and cluster specialist
Copyright © 2014, Oracle and/or its affiliates. All rights reserved.

Safe Harbor Statement
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 decisions. The development, release, and timing of any features or
functionality described for Oracle’s products remains at the sole discretion of Oracle.

History of MySQL

Major milestones in history
• 2001 MySQL 3.23 GA
• 2004 MySQL 4.0 InnoDB new SE
• 2005 MySQL 5.0 GA and Oracle acquired Innobase OY
• 2008 SUN acquired MySQL AB for $1 Billion
• 2010 Oracle acquires SUN
• 2010 MySQL 5.5 GA “InnoDB default”
• 2013 MySQL 5.6 GA “Best release ever”

Open Source LAMP Stack
Operating
System
Application
Server
Database
Scripting
L
A
M
P

World’s Most Popular Open Source Database
• Over 15 million estimated installations
• Used by 9 of top 10 web sites in the world
• Embedded by 8 of the top 10 ISVs
• #1 database in the Cloud
• Integrated w/Hadoop in 80% of Big Data platforms
• Facebook: 175K fans, +35% YoY Growth
• Twitter: 28K followers, +67% YoY Growth
• Numerous Awards: Linux Journal, CRN, PHP Architect...

Cloud
Web & Enterprise OEM & ISVs
Industry Leaders Rely on MySQL

MySQL Powers The Web
Over 500 million Tweets/day. 143,200 Tweets/sec in Aug 2013
”Many petabytes” of data. 11.2 Million Row changes & 2.5 billion
rows read /sec handled in MySQL
6 billion hours of video watched each month
Globally-distributed database with 100 terabytes of user-related
data based on MySQL Cluster

MySQL Arcitecture

MyISAM
• Default Storage Engine before MySQL 5.5
• Table level locking
• Small footprint on disk
• Read Only during backups
• GIS and FTS indexing

InnoDB
• Default Storage Engine since MySQL 5.5
• ACID compliant transactions, MVCC
• Row level locking
• Clustered primary key index
• Foreign keys and and native partitioning
• Online backup and DDL operations
• SQL and NoSQL access to data

MySQL Cluster aka NDB
• Clustered database
• ACID compliant
• Row level locking
• Shards data automatically, transparent for application
• Foreign keys and and native partitioning
• Online backup and DDL operations
• SQL and NoSQL access to data
• Max 2TB

MySQL 5.5
• InnoDB default SE
• InnoDB multiple BP, splitting and removing mutexes
• InnoDB 10X improvement in InnoDB recovery speed
• Replication semi-synchronous replication
• Replication heartbeat
• Partitioning on more data types and more partition functions
• Performance Schema released

MySQL 5.6 GA

•Scales to 48 CPU Threads
•Up to 230% performance gain over MySQL 5.5
IMPROVED PERFORMANCE AND SCALABILITY
•Better transactional throughput and availability
IMPROVED INNODB
•Faster query execution and diagnostics for query tuning and debugging
IMPROVED OPTIMIZER
•Higher performance, availability and data integrity
IMPROVED REPLICATION
•Better Instrumentation, User/Application level statistics and monitoring
IMPROVED PERFORMANCE SCHEMA
•Fast, Key Value access with full ACID compliance, better developer agility
New! NoSQL ACCESS TO INNODB
MySQL 5.6: GA 2 Years Ago – Best GA Ever

 Several internal improvements (e.g. Split kernel mutex, efficient Buffer Pool
Flushing, more...)
 Optimized for Read Only workloads
 Persistent Optimizer Statistics
- Increased plan stability, accurate statistics
- Better user control, automatic/manual
 SSD Optimizations
- 4, 8k page sizes
- .ibd files outside of MySQL data dir
- separate tablespaces for undo log
MySQL 5.6: InnoDB
Better Performance, Scalability
dev.mysql.com/doc/refman/5.6/en/innodb-performance.html

 Optimized for Read Only workloads
- Highly concurrent, read intensive web apps
- Enables developer control of read only transactional overhead
MySQL 5.6: InnoDB
Better Performance, Scalability
SET autocommit = 1;
SELECT c FROM sbtest WHERE id=N;
SET autocommit = 0;
START TRANSACTION READ ONLY;
SELECT c FROM sbtest WHERE id=N;
COMMIT;
On by default
Developer controlled
dev.mysql.com/doc/refman/5.6/en/innodb-performance.html#innodb-performance-ro-txn

MySQL 5.6.7
MySQL 5.6 SysBench Benchmarks
Scales to 48 CPU Threads
Oracle Linux 6
Intel(R) Xeon(R) E7540 x86_64
MySQL leveraging:
- 48 of 96 available CPU threads
- 2 GHz, 512GB RAM
0
2 000
4 000
6 000
8 000
10 000
12 000
12 24 36 48
TransactionsperSecond
CPU Threads
MySQL 5.6 Read Write (Linux)

MySQL 5.6.7
MySQL 5.6 SysBench Benchmarks
Scales to 48 CPU Threads
Oracle Linux 6
Intel(R) Xeon(R) E7540 x86_64
MySQL leveraging:
- 48 of 96 available CPU threads
- 2 GHz, 512GB RAM
0
2 000
4 000
6 000
8 000
10 000
12 000
14 000
16 000
18 000
12 24 36 48
TransactionsperSecond
CPU Threads
MySQL 5.6 Read Only (Linux)

 Online DDL Operations implemented:
- CREATE INDEX
- DROP INDEX
- Change AUTO_INCREMENT value for a column
- ADD/DROP FOREIGN KEY
- Rename COLUMN
- Change ROW FORMAT, KEY_BLOCK_SIZE for a table
- Change COLUMN NULL, NOT_NULL
- Add, drop, reorder COLUMN
MySQL 5.6: InnoDB
Better Availability
dev.mysql.com/doc/refman/5.6/en/innodb-online-ddl.html

 Dump and restore/warm buffer pool
- Shortens warm up times after restart (from hours to minutes)
- At shutdown/startup or manually at any time
- Small footprint on disk - stores tablespace and page IDs only
MySQL 5.6: InnoDB
Better Availability
dev.mysql.com/doc/refman/5.6/en/innodb-performance.html
Example:
Dump current state of Buffer Pool at shutdown:
mysql> SET innodb_buffer_pool_dump_at_shutdown=ON;
Load Buffer Pool from dump at startup:
mysql> SET innodb_buffer_pool_load_at_startup=ON;

 FULLTEXT indexes on InnoDB tables
 Keys on text-based content
 Speeds up searches for words, phrases
 Fully transactional, fast look up
 Natural language/Boolean modes, proximity
search, relevance ranking
MySQL 5.6: InnoDB
Full Text Search create table quotes
( id int unsigned
auto_increment primary
key
,author varchar(64)
, quote varchar(4000)
, source varchar(64)
, fulltext(quote)
) engine=innodb;
select author as “Apple" from quotes
where match(quote) against (‘apple' in natural language mode);
dev.mysql.com/doc/refman/5.6/en/innodb-table-and-index.html#innodb-fulltext-index

 Transportable Tablespaces
- Enables export/import of tables between running MySQL instances
- Faster than mysqldump (via portable .ibd files)
MySQL 5.6: InnoDB
More Flexibility
CREATE TABLE t(c1 INT) engine=InnoDB;
FLUSH TABLE t FOR EXPORT; -- flush table and create the meta data file
$innodb_data_home_dir/test/t.cfg
UNLOCK TABLES;
Export:
Import:
CREATE TABLE t(c1 INT) engine=InnoDB; -- if it doesn't already exist
ALTER TABLE t DISCARD TABLESPACE;
-- The user must stop all updates on the tables, prior to the IMPORT
ALTER TABLE t IMPORT TABLESPACE;

 Key-value access to InnoDB
- Bypasses SQL Parsing
 SQL for rich queries, JOINs, FKs,
etc.
 Implemented via:
- Memcached plug-in to mysqld
- Memcached mapped to native
InnoDB API
- Use existing Memcached clients
- Shared process for ultra-low latency
MySQL 5.6: InnoDB
More Flexibility - Memcached API
dev.mysql.com/doc/refman/5.6/en/innodb-memcached.html

Performance
Up to 9x Higher “SET / INSERT” Throughput
0
10000
20000
30000
40000
50000
60000
70000
80000
8 32 128 512
TPS
Client Connections
MySQL 5.6: NoSQL Benchmarking
Memcached API
SQL
Customized Java-based App Test Harness
Oracle Linux 6
Intel(R) Xeon(R) 2.0 x86_64
MySQL leveraging:
- 8 available CPU threads
- 2 GHz, 16GB RAM
blogs.oracle.com/mysqlinnodb/entry/new_enhancements_for_innodb_memcached

 For Developers
- No more re-writing legacy application or packaged subqueries into joins
 Performance
- DBT 3 Query #18 benchmark:
- Execution time drops from DAYS to seconds
MySQL 5.6: Optimizer
Subquery Optimizations
SELECT title FROM film WHERE film_id IN
(SELECT film_id FROM film_actor
GROUP BY film_id HAVING count(*) > 12);
dev.mysql.com/doc/refman/5.6/en/subquery-optimization.html

 EXPLAIN
- INSERT, UPDATE, and DELETE
- Structured EXPLAIN output
 Optimizer Traces
Better Diagnostics
SET SESSION OPTIMIZER_TRACE=‘enabled=on’;
SELECT (SELECT 1 FROM t6 WHERE d = c)
AS RESULT FROM t5;
SELECT * FROM information_schema.OPTIMIZER_TRACE;
"records_estimation": [
{
"database": "test",
"table": "t6",
"range_analysis": {
"table_scan": {
"records": 2,
"cost": 4.5034
},
"potential_range_indices": [
{
"index": "d",
"usable": true,
"key_parts": [
"d"
]
}
],
"best_covering_index_scan": {
"index": "d",
"cost": 1.4233,
"chosen": true
},
dev.mysql.com/doc/refman/5.6/en/explain.html
dev.mysql.com/doc/internals/en/optimizer-tracing.html

•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
•Time Delayed Replication
•Remote Binlog Backup
DEV/OPS AGILITY
MySQL 5.6: Replication
dev.mysql.com/tech-
resources/articles/mysql-5.6-
replication.html
 New default and
Developer controlled
improvements

 Simple to track & compare replication across the cluster
- Unique identifier for each transaction written to the Binlog
 Automatically identify the most up-to-date slave for failover
 Deploy n-tier replication hierarchies
Master
GTID=123456
GTID=123456
GTID=123456 GTID=123456
MySQL 5.6: Replication
New! Global Transaction Ids
dev.mysql.com/doc/refman/5.6/en/replication-gtids.html

MySQL 5.6 Performance Schema
We’ve been busy!
New Instrumentation
• Statements/Stages
• Table and Index I/O
• Table locks
• Users/Hosts/Accounts
• Network I/O
New Features
 Show contents of Host cache
 New Summary tables
 Easier configuration
- Start up defaults in my.cnf
- Auto tune
 Reduced overhead
 On by default
[mysqld]
performance_schema=on
dev.mysql.com/doc/refman/5.6/en/performance-schema.html

 Major overhaul of password handling
- Provide alternatives to showing passwords in plain text
- Assess/Enforce password strength policies
- Enforce new password at next login
- Stronger password hashing
MySQL 5.6: Security

MySQL Roadmap

5.7 DMR 1
April 2013
DMR 2
Sep 2013
DMR 3
Dec 2013
DMR 4
March 2014
5.6 GA
Feb 2013
Raising the Bar: MySQL 5.7
DMR 5 (5.7.5)
OOW 2014

MySQL 5.7: DMR 4
MySQL 5.7 builds on MySQL 5.6 by improving
• InnoDB for better transactional throughput, availability, IO
• Replication for better scalability and availability
• Utilities for dev/ops automation
• Performance Schema for better performance metrics
• Optimizer for better EXPLAINing, query performance, enhanced
buffering and partition optimization
• Connecting at higher rates, improve session efficiency

Sysbench Point Select
Intel(R) Xeon(R) CPU X7560 x86_64
5 sockets x 8 cores-HT (80 CPU threads)
2.27GHz, 256G RAM
Oracle Linux 6.5
2X Faster than MySQL 5.6
Over 3X Faster than MySQL 5.5
630,000 QPS

EXPLAIN on Running Queries
• Problem: a statement in a session is taking a long time to complete
• New Option: run
– EXPLAIN FOR CONNECTION from another session
– Discover the root cause of the delay
– Optimize the statement
EXPLAIN [FORMAT=(JSON|TRADITIONAL)] FOR CONNECTION <id>;

MySQL 5.7: Optimizer Cost Model
 New cost model API
 Allows storage engines to provide accurate and dynamic cost estimates for key
lookups, table scans, range scans, etc…
‒ Enables future support of additional factors
 Whether the data is in RAM, SSD, HDD
 Lays the groundwork for making costs configurable
‒ Based on your hardware performance characteristics
 Improves records per key estimates
 Cost values included in JSON Explain output

MySQL 5.7: InnoDB
Many Improvements
• Improved InnoDB Online Alter Table
– Online Rename Index, Online Change Varchar
• Enhanced FusionIO Integration
– Doublewrite buffer automatically disabled when DirectFS detected
• Parallel “Dirty Page” Flushing
– Higher throughput, performance, and scalability
• Partitions – support for Transportable Tablespaces (TTS)
– TTS support for individual partitions

MySQL 5.7: Server-side Statement Timeouts
• Server Side statement timeouts
– Global for server, per session, or for individual SELECT statements
• Based on contribution from Davi Arnaut
– Expanded to Windows and Solaris, restricted by removing USER option
SELECT MAX_STATEMENT_TIME = 109 * FROM my_table;

MySQL GIS
• InnoDB Spatial Index Support
– Optimized R-tree indexes
• Boost.Geometry integration
– Improved accuracy
– Support for additional types, combinations, and functions
– Improved performance
– Full Open Geospatial Consortium compliance
– Enhancements to Boost.Geometry pushed upstream
labs.mysql.com

MySQL 5.7.4: Replication
Better Performance, Improved Usability and Enhanced HA
• Higher throughput
– Slave : applies transactions in parallel even within same database
– Master: better synchronization between replication-user sessions
• DMR4 improves master performance for Semi-synchronous Replication
• Lossless Replication through enhanced Semi-sync
• Performance Schema tables for monitoring slave
• Dynamic Replication Filters: change filters on the fly

Multi-Source Replication
• Consolidate updates from multiple
Masters into one Slave
– Consolidated view of all shards
– More flexible topologies
– Centralized point for backups, easier reporting
• Compatible with Semi-Synchronous
Replication & enhanced MTS
• Master-specific slave filters planned for
GA
• Application must keep data sets disjoint
between sources
Binlog
Master 1
Binlog
Master 2
…
…
Binlog
Master N
IO 1
Relay 1
Coordinator
W1 W2 … WX
IO 2
Relay 2
Coordinator
W1 W2 … WX
…
…
Coordinator
W1 W2 … WX
IO N
Relay N
Coordinator
W1 W2 … WX
Slave
labs.mysql.com

MySQL Future
Focus on Web, Cloud, Embedded
• Architecture Improvements
– Improved Modularity
– Refactoring Optimizer, Parser, Runtime
• Web Scale Performance & Scalability
– Optimizer Cost Model
– InnoDB & Replication Improvements
• Management & Security
– Data Dictionary
– MySQL Enterprise Encryption
– Oracle Enterprise Manager for MySQL

MySQL High Availability

MySQL
Replication
MySQL
Fabric
DRBD
Windows/S
olaris/Clust
erware
Clustering
or Oracle
VM
MySQL
Cluster
MySQL HA Solutions
9 9 . 9 9 9 %

Mapping Uptime to Availability
Replicated
Systems
• SPs & Line of
Business
Clustered &
Virtualized
Systems
• Web & Cloud
Services
Shared
Nothing,
Geo-
Replicated
Clusters
• eCommerce,
Telecoms &
Military
9 9 . 9 9 9 %
Cost&Complexity

MySQL Fabric 1.5
• High Availability
– Server monitoring with auto-promotion and
transparent application failover
• Optionally scale-out through sharding
– Application provides shard key
– Range or Hash
– Tools for resharding
– Global updates & tables
• Fabric-aware connectors rather than
proxy: Python, Java, PHP, .NET, C (labs)
– Lower latency, bottleneck-free
• Server provisioning using OpenStack etc.
High Availability + Sharding-Based Scale-out
MySQL Fabric
Connector
Application
Read-slaves
SQL
HA group
Read-slaves
HA group
Connector
Application

• Oracle Clusterware unifies servers
in a server farm to form a cluster
– At the core of Oracle RAC
• Oracle Cluster 12c includes MySQL
Server 5.6 agent
• Planned migration and failover of
MySQL database
– Hidden from the application
MySQL on Oracle Clusterware

• Native Windows HA clustering with
MySQL
• Quorum (3rd vote), data (InnoDB +
schema) & binaries (optional)
stored in shared storage (iSCSI &
FCAL)
• Loss of service = couple of seconds
+ InnoDB recovery time
• Cluster managed through MS
Failover Cluster Management snap-
in GUI
Windows Server Failover Clustering
App
Virtual IP
Data Bin
Slave App App
Vote

• Kernel based heartbeating and
monitoring
• SPARC and x86. Solaris
Virtualization-aware
• MySQL agent included with Oracle
Solaris Cluster
• Learn more:
http://www.oracle.com/technetwo
rk/server-storage/solaris-
cluster/overview/index.html
Oracle Solaris Clustering

• Based on distributed storage
– NOT physical shared storage
• Synchronous replication at block
device level eliminates risk of data
loss
• Open source, mature & proven
• Certified and supported by Oracle
Oracle Linux and DRBD Stack
Hosts
Oracle Linux Oracle Linux
Active Standby
Corosync
Pacemaker
Cluster s/w
DRBD DRBD
MySQL MySQL
Application
VIP
sync
Services

• Pre-Installed & Pre-Configured
• Full Integration & QA Testing
• Single Point of Support
Oracle VM Template for MySQL
Oracle VM Servers
Oracle VM Server Pool
ocfs2
Oracle VM
Manager
iSCSI / FC-AL
Oracle VM
Secure Live
Migration (SSL)
Oracle VM
Automatic Fault
Detection &
Recovery

MySQL Cluster Overview
• Auto-Sharding, Multi-Master
• ACID Compliant, OLTP + Real-Time Analytics
HIGH SCALE, READS +
WRITES
• Shared nothing, no Single Point of Failure
• Self Healing + On-Line Operations
99.999% AVAILABILITY
• In-Memory Optimization + Disk-Data
• Predictable Low-Latency, Bounded Access Time
REAL-TIME
• Key/Value + Complex, Relational Queries
• SQL + Memcached + JavaScript + Java + HTTP/REST & C++
SQL + NoSQL
• Open Source + Commercial Editions
• Commodity hardware + Management, Monitoring Tools
LOW TCO

MySQL Cluster Architecture
MySQL Cluster Data Nodes
Clients
Application Layer
Data Layer

MySQL Cluster Scaling
Clients
Application Layer
Data Layer

MySQL Cluster HA
Clients
Application Layer
Data Layer

Who’s Using MySQL Cluster?

Oracle MySQL HA & Scaling Solutions
MySQL
Replication
MySQL
Fabric
Oracle VM
Template
Oracle
Clusterware
Solaris
Cluster
Windows
Cluster
DRBD
MySQL
Cluster
App Auto-Failover ✖ ✔ ✔ ✔ ✔ ✔ ✔ ✔
Data Layer Auto-Failover ✖ ✔ ✔ ✔ ✔ ✔ ✔ ✔
Zero Data Loss MySQL 5.7 MySQL 5.7 ✔ ✔ ✔ ✔ ✔ ✔
Platform Support All All Linux Linux Solaris Windows Linux All
Clustering Mode
Master +
Slaves
Master +
Slaves
Active/Pas
sive
Active/Passi
ve
Active/P
assive
Active/Pas
sive
Active/P
assive
Multi-
Master
Failover Time N/A Secs Secs + Secs + Secs + Secs + Secs + < 1 Sec
Scale-out Reads ✔ ✖ ✖ ✖ ✖ ✖ ✔
Cross-shard operations N/A ✖ N/A N/A N/A N/A N/A ✔
Transparent routing ✖ For HA ✔ ✔ ✔ ✔ ✔ ✔
Shared Nothing ✔ ✔ ✖ ✖ ✖ ✖ ✔ ✔
Storage Engine InnoDB+ InnoDB+ InnoDB+ InnoDB+ InnoDB+ InnoDB+ InnoDB+ NDB
Single Vendor Support ✔ ✔ ✔ ✔ ✔ ✖ ✔ ✔

Best Choice for Next Generation Web & Cloud Applications
World’s Most Popular Open
Source Database
Leading Open Source
Database for Web
Applications
#1 Open Source Database
in the Cloud
Integrated with Hadoop in
Big Data Platforms
Strong MySQL Momentum

Getting Started
MySQL Enterprise Edition
• http://www.mysql.com/news-and-events/web-seminars/why-relying-on-mysql-enterprise-edition/
• http://www.mysql.com/news-and-events/web-seminars/the-mysql-sys-schema/
MySQL Security and Upgrading to 5.6
• http://www.mysql.com/news-and-events/web-seminars/mysql-security-whats-new-in-mysql-5-7-best-practices/
• http://www.mysql.com/news-and-events/web-seminars/upgrading-to-mysql-5-6-best-practices/
MySQL Availability
• http://www.mysql.com/why-mysql/white-papers/#en-22-16
• http://www.mysql.com/news-and-events/on-demand-webinars/#en-20-16

Thank You!

MySQL 5.6, news in 5.7 and our HA options

Recommended

More Related Content

What's hot (20)

Viewers also liked (9)

Similar to MySQL 5.6, news in 5.7 and our HA options (20)

More from Ted Wennmark (8)

Recently uploaded (20)

MySQL 5.6, news in 5.7 and our HA options