Faster Than Alter – Less Downtime Chris Schneider.

Faster Than Alter – Less Downtime

Chris Schneider

Overview

Some problems

Alter table

Load Data INFILE

Fragmentation

De-Fragmentation

Some tools to consider

Use of slave servers

Real world application

- Schema change

- Defragmentation

Some Problems

DDL can be a pain in a working system

Need to add or remove a column

Need to add or remove a clustered or secondary key

Running into performance degradation

- You’ve tried tuning the my.cnf but still need better performance

Alter Table

Alter table is very useful but can be unmanageable with large data sets

Everyone in here should about alter table!!!!!!!

Classic INNODB

- ALTERs will run a table rebuild

- Innodb Table space will grow

Innodb Plugin (Bonus)

- Table space will be rebuilt in place WHEN:

-- Dropping a secondary index

-- Adding a secondary index

Load Data Infile

Load Data Infile is very fast and pretty easy to work with

Most people in here should know about LOAD DATA INFILE!!!

You should also know about SELECT INTO OUTFILE

Also the tab delimited mysqldump

Working with TSV and CSV files

- Make sure when you export you do it in PK order

Fragmentation

Happens over time

Write activity creates holes and degrades performance

You did not load your data set initially in PK order

Can happen by accident

- Update table set column1 = 2 where column1 = 1;

Defragmentation

You will see performance gains from a dump out and reload

- 30% - 80% increase in overall throughput

For MyISAM and INNODB setups you will regain disk space

Upper management will be pleased with the sudden increase in database performance

Your customers will also be pleased, we HOPE!

Depending on your server count and setup you might only have to do this twice a year or less

Defragmentation (Cont)

Tools to Consider

mk-parallel-dump

- Dump tables in parallel

mk-parallel-restore

- Load tables in parallel

mk-fifo-split

- Split files and pipe lines to a fifo without really splitting

Select into outfile

LOAD data infile

MySQL Tab delimited dump

Using Slave Servers

Slave servers can be used to run alters or run an dump and reload of data

Some Considerations

- Do you have a slave?

- Do you have a good failover mechanism

- Semi-Automated

- Full-Automation

The General Idea:

1. ON the slave, Stop slave, Select INTO OUTFILE, LOAD DATA INFILE

2. Fail over from the master to the slave

3. Repeat step 1 on OLD MASTER

Using Slave Servers (Cont)

Using Slave Servers (Cont)

Using Slave Servers (Cont)

Real World Applications

Example 1: NEW COLUMN and INDEX needed

I have a system with 10, two node clusters, one master and one slave in each cluster

Each cluster has 10 databases on it with an average dataset of 70GB

The master servers cannot go down or be locked or an extended period of time. In this case I was given a 10 minute window (off hours) for a maintenance window

Server Specs

Dell 2850

16GB RAM

Dual Quad – Intel 2.50GHz

8 x 300GB 10K SATA RAID 10

Real World Applications (CONT)

Example 2: Defragmentation

Did you know that it takes 2 years to paint the Golden Gate bridge?

I have a system with thousands of servers that have been running a 60/40 (READ/WRITE) ratio for years

Each system is INNODB and has, on average, 170GB of data and index

Server Specs

HP DL380 G5

32GB RAM

Dual Quad – Intel Xeon L5420 @ 2.50GHz

8x146GB 15K SAS

Two RAID sets

- 2 disk mirror for OS

- 6 disk RAID 10 for DATA

FYI and Questions

Slides will go up on http://everythingmysql.com

Questions?