Laracon EU 2018: OMG MySQL 8.0 is out! are we there yet?

OMG MySQL 8.0 is out! Are
we there yet?
Gabriela Ferrara
@gabidavila
http://gabriela.io | gabi.fyi/feedback

@gabidavila
Gabi
!
Developer Advocate

@gabidavila
Cloud SQL
PostgreSQLMySQL

@gabidavila
Are we there yet?
🤔
● Created to handle from up to
10 to 100M rows or around
100MB/table
● Now supports terabyte-sized
databases
● Supports parts of SQL
standards as new as SQL 2016
● But... some stuff from SQL
2003 just became available
(Window Functions, CTEs)
● What can we go from here?

@gabidavila
What Happened to 6.0 and 7.0?

@gabidavila
Topics
● New Defaults & Variables
● InnoDB Performance
● Window Functions
● CTE - Common Table Expressions

@gabidavila
New Defaults & Variables

@gabidavila
New default Charset
● Default:
● 5.7: latin1
● 8.0: utf8mb4
● Improvements:
● ➕ Mathematical Equations 𝑒=𝑚·𝑐²
● 😁 🙄 &
● & more SMP (Supplementary Multilingual Plane) Characters

@gabidavila
New default Collation
● utf8mb4_0900_ai_ci
● UTF-8 version 9.0 support
● Accent Insensitive
● Case Insensitive
● No more 🍣 = 🍺 bug
● Caused by utf8mb4_general_ci or utf8mb4_unicode_ci
More information on how collations behave here.

@gabidavila
Other defaults & variables
● Binary log (log_bin) is enabled by default
● SHA-2 for authentication
● Mandatory default value for TIMESTAMP NOT NULL
● New variable to dedicated servers (default OFF),
innodb_dedicated_server=ON , controls dynamically:
● innodb_buffer_pool_size
● innodb_log_file_size
● innodb_flush_method

@gabidavila
InnoDB Performance

@gabidavila
Transaction Scheduling
RDBMS usually works on a FIFO architecture
Transaction A
Transaction B
Transaction C
Start
End

@gabidavila
CATS
First database in the world to
implement

@gabidavila
Contention Aware Transaction Scheduling

@gabidavila
CATS vs FIFO | Transactions per Second (TPS)
0
5500
11000
16500
22000
32 64 128 256 512
FIFO CATS
DataextractedfromMySQLServerTeamwebsite.
Transactions per Second x # Clients
Higher is better

@gabidavila
CATS vs FIFO | Latency
0
45
90
135
180
32 64 128 256 512
FIFO CATS
DataextractedfromMySQLServerTeamwebsite.
Mean Latency x # Clients
Lower is better

@gabidavila
ALGORITHM=INSTANT
● If possible, the default is INSTANT, else it will be INPLACE
● Only metadata changed on the Data Dictionary
● Supported on:
● Change index
● Rename Table
● SET/DROP DEFAULT
● MODIFY COLUMN
● ADD/DROP Virtual Columns
● ADD COLUMN (non-generated)
🐞 Tiny bug if you are migrating from 8.0.11

@gabidavila
DESCRIBE `lines`;
type null extra
id int(11) NO AUTO_INCREMENT
number int(11) YES
timpestamp_in_ms timestamp YES
character_id int(11) NO
episode_id int(11) NO
location_id int(11) NO
character_line text YES
word_count int(11) YES

@gabidavila
Invisible Indexes
● Not used by the optimizer
● Visible by default, to create an invisible index:
ALTER TABLE lines
ADD INDEX ix_word_count (word_count) INVISIBLE;
ALTER TABLE lines ALTER INDEX ix_word_count INVISIBLE;
ALTER TABLE lines ALTER INDEX ix_word_count VISIBLE;
● Toggle visibility:

@gabidavila
Now imagine if an Eloquent migration…

@gabidavila
Laravel
Taylor will be waiting your pull request!

@gabidavila
Invisible Indexes
SELECT *
FROM lines
WHERE word_count > 25
ORDER BY word_count DESC;
Invisible
Query cost: 59175.37
Visible
Query cost: 2847.86

@gabidavila
Descending Indexes
Up to 5.7
● Syntax allowed ASC or DESC when defining an index
However...
An index_col_name specification can end with ASC or DESC. These keywords are
permitted for future extensions for specifying ascending or descending index value
storage. Currently, they are parsed but ignored; index values are always stored in
ascending order.

@gabidavila
MySQL 8.0 has it!!!

@gabidavila
Descending Indexes
● No longer ignored and forcibly created as ASC
ALTER TABLE lines
ADD INDEX ix_word_count (word_count DESC);

@gabidavila
Descending Indexes
Can be scanned as intended or backwards
SELECT *
FROM lines
ORDER BY word_count DESC;
-- OR WITH THE SAME COST
SELECT *
FROM lines
ORDER BY word_count;

@gabidavila
More cool features!

@gabidavila
What they do?
● Allows to analyze the rows of a given result set
● Can behave like a GROUP BY without changing the result set
● Allows you to use a frame to "peek" OVER a PARTITION of a window

@gabidavila
Window Functions
● Examples:
● Enumerate rows - ROW_NUMBER()
● Show Aggregated sums - SUM()
● Rank results - RANK(), DENSE_RANK()
● Look at neighboring rows - LEAD(), LAG()

@gabidavila
DESCRIBE èpisodes`;
CREATE TABLE èpisodes` (
ìd` int(11) NOT NULL,
`title` varchar(255),
èpisode_date` date,
`season` int(11),
`number_in_season` int(11),
`number_in_series` int(11),
`million_viewers` decimal(10, 2),
ìmdb_rating` decimal(10, 2),
ìmdb_votes` decimal(10, 2),
ìmage_url` text,
`video_url` text,
PRIMARY KEY (ìd`)
) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4;

@gabidavila
SELECT … FROM `episodes` LIMIT 10;
+----+--------------+-----------------------------------+--------+------------------+-----------------+
| id | episode_date | title | season | number_in_season | million_viewers |
+----+--------------+-----------------------------------+--------+------------------+-----------------+
| 1 | 1989-12-17 | Simpsons Roasting on an Open Fire | 1 | 1 | 26.70 |
| 2 | 1990-01-14 | Bart the Genius | 1 | 2 | 24.50 |
| 3 | 1990-01-21 | Homer's Odyssey | 1 | 3 | 27.50 |
| 4 | 1990-01-28 | There's No Disgrace Like Home | 1 | 4 | 20.20 |
| 5 | 1990-02-04 | Bart the General | 1 | 5 | 27.10 |
| 6 | 1990-02-11 | Moaning Lisa | 1 | 6 | 27.40 |
| 7 | 1990-02-18 | The Call of the Simpsons | 1 | 7 | 27.60 |
| 8 | 1990-02-25 | The Telltale Head | 1 | 8 | 28.00 |
| 9 | 1990-03-18 | Life on the Fast Lane | 1 | 9 | 33.50 |
| 10 | 1990-03-25 | Homer's Night Out | 1 | 10 | 30.30 |
+----+--------------+-----------------------------------+--------+------------------+-----------------+
10 rows in set (0.00 sec)

@gabidavila
Previous and Next episodes | LAG and LEAD
SELECT
id,
title,
LAG(episode_date, 1) OVER (ORDER BY episode_date) AS previous_episode,
episode_date,
LEAD(episode_date, 1) OVER (ORDER BY episode_date) AS next_episode,
million_viewers
FROM episodes;

@gabidavila
SELECT … FROM `episodes` LIMIT 10;
+----+-----------------------------------+------------------+--------------+--------------+-----------------+
| id | title | previous_episode | episode_date | next_episode | million_viewers |
+----+-----------------------------------+------------------+--------------+--------------+-----------------+
| 1 | Simpsons Roasting on an Open Fire | NULL | 1989-12-17 | 1990-01-14 | 26.70 |
| 2 | Bart the Genius | 1989-12-17 | 1990-01-14 | 1990-01-21 | 24.50 |
| 3 | Homer's Odyssey | 1990-01-14 | 1990-01-21 | 1990-01-28 | 27.50 |
| 4 | There's No Disgrace Like Home | 1990-01-21 | 1990-01-28 | 1990-02-04 | 20.20 |
| 5 | Bart the General | 1990-01-28 | 1990-02-04 | 1990-02-11 | 27.10 |
| 6 | Moaning Lisa | 1990-02-04 | 1990-02-11 | 1990-02-18 | 27.40 |
| 7 | The Call of the Simpsons | 1990-02-11 | 1990-02-18 | 1990-02-25 | 27.60 |
| 8 | The Telltale Head | 1990-02-18 | 1990-02-25 | 1990-03-18 | 28.00 |
| 9 | Life on the Fast Lane | 1990-02-25 | 1990-03-18 | 1990-03-25 | 33.50 |
| 10 | Homer's Night Out | 1990-03-18 | 1990-03-25 | 1990-04-15 | 30.30 |
+----+-----------------------------------+------------------+--------------+--------------+-----------------+

@gabidavila
Break down
LAG(episode_date, 1) OVER (ORDER BY episode_date)
windowfunction
column
# rows preceding

@gabidavila
Repetition?
SELECT
id,
title,
LAG(episode_date, 1) OVER (ORDER BY episode_date) AS previous_episode,
episode_date,
LEAD(episode_date, 1) OVER (ORDER BY episode_date) AS next_episode,
million_viewers
FROM episodes;

@gabidavila
SELECT
id,
title,
LAG(episode_date, 1) OVER (dates) AS previous_episode,
episode_date,
LEAD(episode_date, 1) OVER (dates) AS next_episode,
million_viewers
FROM episodes
WINDOW dates AS (ORDER BY episode_date);
Named Windows!

@gabidavila
DEMO
Rank episodes by million_viewers

@gabidavila
CTE 
Common Table Expressions

@gabidavila
Common Table Expressions
● Similar to CREATE [TEMPORARY] TABLE
● Doesn’t need CREATE privilege
● Can reference other CTEs (if those are already defined)
● Can be recursive
● Easier to read

@gabidavila
Rank episodes by million_viewers
SELECT
id,
season,
number_in_season AS '# in season',
title,
episode_date,
million_viewers,
DENSE_RANK() OVER (season_viewers) AS episode_rank
FROM episodes
WINDOW season_viewers
AS (PARTITION BY season ORDER BY million_viewers DESC)

@gabidavila
DEMO
Filter ranked episodes

@gabidavila
Recursive CTE
● Useful with hierarchical data
● The Recipe is:
● Base query comes first
● Second query comes after a UNION statement
● And the stop condition should be on the recursive call

@gabidavila
Fibonacci
WITH RECURSIVE fibonacci(recursion_level, fibonacci_number, next_number)
AS (
# Base Case
SELECT 0 AS recursion_level,
0 AS fibonacci_number,
1 AS next_number
UNION ALL
# Recursion query
SELECT recursion_level + 1,
next_number,
fibonacci_number + next_number
FROM fibonacci
# Stopping condition
WHERE
recursion_level < 10
)
SELECT * FROM fibonacci;

@gabidavila
Fibonacci
+-----------------+------------------+-------------+
| recursion_level | fibonacci_number | next_number |
+-----------------+------------------+-------------+
| 0 | 0 | 1 |
| 1 | 1 | 1 |
| 2 | 1 | 2 |
| 3 | 2 | 3 |
| 4 | 3 | 5 |
| 5 | 5 | 8 |
| 6 | 8 | 13 |
| 7 | 13 | 21 |
| 8 | 21 | 34 |
| 9 | 34 | 55 |
| 10 | 55 | 89 |
+-----------------+------------------+-------------+
Base case
Recursive Query

@gabidavila
Other MySQL 8.0 Bits
● Users now can have ROLES
● JSON partial update
● Better Geospatial support
● Better UUID support
● Use MySQL without SQL (through MySQL Shell or X-DevApi)

@gabidavila
Thank you!
Please give me feedback on my talk!
gabi.fyi/feedback

Laracon EU 2018: OMG MySQL 8.0 is out! are we there yet?

More Related Content

What's hot (10)

Similar to Laracon EU 2018: OMG MySQL 8.0 is out! are we there yet? (20)

More from Gabriela Ferrara (16)

Recently uploaded (20)

Laracon EU 2018: OMG MySQL 8.0 is out! are we there yet?