How indexes work in relational databases Tokyo Rubyist Meetup Feb. 10th, 2017

Who am I doublemarket ● Technical support for a RoR based package software ● Worked as MySQL DBA ● Hobbies ○ Translation - Yakst https://yakst.com/ja ○ Creating twitter bots ● Twitter : dblmkt

SQL Performance Explained ● How indexes work ○ Multiple databases : Oracle, MySQL, PostgreSQL, etc. ○ Multiple languages : PHP, Java, etc. ● Written by Markus Winand ● Available in EN, JP, GE, FR, SP ○ Japanese ver. translator = me ● Buy at http://sql-performance-explained.com/ ○ JP : http://sql-performance-explained.jp/ ● Free Web Edition: http://use-the-index-luke.com/

What I’m going to talk ● Why you need indexs (B-tree index) ● How indexes work in general ● Pros and Cons of indexes

What I’m NOT going to talk about ● How to write SQL queries (w/ Ruby on Rails) ○ Buy and read “SQL Performance Explanined” ;-) ○ Read AR manual ● How to tune SQL queries ○ Depends on a DBMS you’re using

What is good points of SQL?

What you want How you get

What you want How you get When writing a SQL query

What you want How you get When getting a result quickly or efficiently

What you want How you get

Developers need to know how to get data from storage ≒ how to use indexes

What is index?

Indexes in a relational database Same as an index of a book ● To get data quickly and efficiently ● Redundant, just pointing a piece of data stored somewhere ● Sorted by a specific order Different from an index of a book ● Updated frequently

Ice cream Hand Nose Girl X’mas Dog Man Rain Fish Cat Orange Pen Egg Boy Van Water Sugar Apple Queen Tree Yellow Kite Lamp Jet Zoo Umbrella 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 Data

Why don’t you scan all records everytime? ● No, Inefficient ○ Especially data is not on memory ● Time to find a record will be longer when data is getting bigger

A F D G C J B I H E

Why don’t you sort records like a dictionary? ● No, it’s inefficient.

A B C E F G H I J D

A B C E F G H I D J

A B C E F G H D J I

A B C E F G D J I H

A B C E F D J I H G

A B C E D J I H G F

A B C D J I H G F E

A B C D J I H G F E

Use the index, Luke!

Index ≒ B-tree index (in this talk)

How the (B-tree) index works B-tree index ● The most popular index structure for relational databases ● Balanced tree, not binary tree Doubly linked list ● Can change an order by changing pointers ● Can connects leaf nodes

I R Z I J K L M N O P Q R S T E F G H A B C D U V W X Y Z C F U X L R O I Z Ice cream Hand Nose Girl X’mas Dog Man Rain Fish Cat Orange Pen Egg Boy Van Water Sugar Apple Queen Tree Yellow Kite Lamp Jet Zoo Umbrella 18 14 10 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 6 13 9 4 2 1 24 22 23 7 3 11 12 19 8 17 20 26 15 16 5 21 25 (B Tree) Index Data … … … … … … … … … … … … … … … … … … … … … … … … … … …

I R Z I J K L M N O P Q R S T E F G H A B C D U V W X Y Z C F U X L R O I Z 18 14 10 6 13 9 4 2 1 24 22 23 7 3 11 12 19 8 17 20 26 15 16 5 21 25 Root node Branch nodes Leaf nodes Ice cream Hand Nose Girl X’mas Dog Man Rain Fish Cat Orange Pen Egg Boy Van Water Sugar Apple Queen Tree Yellow Kite Lamp Jet Zoo Umbrella 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 … … … … … … … … … … … … … … … … … … … … … … … … … … …

Ice cream Hand Nose Girl X’mas Dog Man Rain Fish Cat Orange Pen Egg Boy Van Water Sugar Apple Queen Tree Yellow Kite Lamp Jet Zoo Umbrella 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 … … … … … … … … … … … … … … … … … … … … … … … … … … … I R Z I J K L M N O P Q R S T E F G H A B C D U V W X Y Z C F U X L R O I Z 18 14 10 6 13 9 4 2 1 24 22 23 7 3 11 12 19 8 17 20 26 15 16 5 21 25 (B Tree) Index Data

Pros of using the (B-tree) index The index ● Reduces a number of disk accesses ● Reduces the amount of disk accesses ● Finds records in a same time when data is getting bigger ● Is smaller than data itself = can fit in memory

Cons of using the index The index ● Consumes storage ● Increases a cost when inserting/deleting/updating records Don’t create too many indices

Summary ● Learn how indexes work on relational databases ● When writing a query ○ Think about how indexes work ○ Think about how the query works when data is getting bigger ● When creating/updating/deleting an index ○ Think about the balance ■ How much the index makes queries fast ■ How much the cost of modifying the index

Learn how indexes work and make your queries fast

http://sql-performance-explained.com http://sql-performance-explained.jp