Diving into MySQL 5.7: advanced features

Diving into MySQL 5.7
Advanced features
Gabriela Ferrara
 
@gabidavila
gabriela.io

@gabidavila
But first...
gabi
Developer Advocate @ Google

@gabidavila
MySQL 5.7 Adoption

@gabidavila
What to expect?
● Online DDL modes
● JSON Data type
● Generated Columns
● sql_mode
● `sys` schema
@gabidavila

Online DDL
changes
On InnoDB
● INPLACE
● COPY
@gabidavila

@gabidavila
New Online DDL Changes on 5.7 | InnoDB Only
In-place | ALGORITHM=INPLACE
● Rename Index
● Add a Virtual Column
● VARCHAR from 1 to 255
@gabidavila
More info 🔗

@gabidavila
New Online DDL Changes on 5.7 | InnoDB Only
Table-copy | ALGORITHM=COPY
● VARCHAR from 255 to 65535
● Drop Stored Column
@gabidavila
More info 🔗

@gabidavila
Table `laps`
Column Type
id INT(10)
athlete VARCHAR(255)
distance (m) INT(10)
time (s) INT(10)
summary JSON

@gabidavila
SELECT * FROM `laps` LIMIT 1;
************************** 1. row *************************** 
id: 1 
athlete: Martha 
distance: 3200 
time: 1800 
summary: {"weight": 55, "nickname": "M", "professional": true} 
1 row in set (0.04 sec)
@gabidavila

@gabidavila
JSON_UNQUOTE + JSON_EXTRACT vs. ->>
@gabidavila
SELECT
athlete,
summary->>'$.nickname' AS nickname,
summary->>'$.professional' AS professional
FROM laps;
😎
SELECT
athlete,
JSON_UNQUOTE(JSON_EXTRACT(summary, '$.nickname')) AS nickname,
JSON_UNQUOTE(JSON_EXTRACT(summary, '$.professional')) AS professional
FROM laps;
😕

@gabidavila
JSON Path
● Access key:
● `$` refers to the document itself
● Keys can contain spaces
● JSON_EXTRACT() and -> are the same
● JSON_UNQUOTE(JSON_EXTRACT()) and ->> are the same

@gabidavila
Types
Virtual
● No disk space
● In-place change operation
● Value generated on demand and
on every BEFORE triggers
Stored
● Uses disk space
● Copy operation
● Updated on every INSERT and
UPDATE
@gabidavila

@gabidavila
Virtual & Stored similarities
● They only know the table domain
● They have a type
● Allows expressions to be used such as:
○ Operators (product * quantity)
○ Built-in functions (YEAR(dob))
○ Literals ("expression", 1)
● Can be indexed

@gabidavila
Virtual & Stored limitations
● Subqueries ARE NOT allowed
● Custom functions ARE NOT supported
● Column rename is not possible
● Non deterministic functions are not supported (i.e. NOW())

@gabidavila
Modified table `laps`
Column Type
id INT(10)
time (s) INT(10)
summary JSON
speed (m/s) DOUBLE
Should it be VIRTUAL or STORED?

@gabidavila
Virtual
Adding the `speed` field to the
`laps` table
Dropping the `speed` field to the
`laps` table
Gist 🔗
ALTER TABLE `laps`
ADD COLUMN speed DOUBLE
GENERATED ALWAYS
AS (`distance` / `time`);
ALTER TABLE `laps`
DROP COLUMN speed;

@gabidavila
Virtual
`laps` table
`laps` table
Gist 🔗
ALTER TABLE `laps`
GENERATED ALWAYS
AS (`distance` / `time`);
-- completed in 82ms
ALTER TABLE `laps`
DROP COLUMN speed;
-- completed in 103ms

@gabidavila
Stored
`laps` table
`laps` table
Gist 🔗
ALTER TABLE `laps`
GENERATED ALWAYS
AS (`distance` / `time`) STORED;
-- 548 rows affected in 778ms
ALTER TABLE `laps`
DROP COLUMN speed;
-- completed in 3s 142ms

Generated
Columns & JSON
Indexing

@gabidavila
SQL Modes | SELECT @@GLOBAL.sql_mode; | MySQL 5.7
On MySQL 5.7:
+-----------------------------------------------------------------------+
| @@GLOBAL.sql_mode |
+-----------------------------------------------------------------------+
| ONLY_FULL_GROUP_BY,STRICT_TRANS_TABLES,NO_ZERO_IN_DATE,NO_ZERO_DATE, | 
| ERROR_FOR_DIVISION_BY_ZERO,NO_AUTO_CREATE_USER,NO_ENGINE_SUBSTITUTION |
+-----------------------------------------------------------------------+
@gabidavila

@gabidavila
SQL Modes | Comparison
On MySQL 5.6:
! STRICT_TRANS_TABLES
! NO_ENGINE_SUBSTITUTION
On MySQL 5.7:
! STRICT_TRANS_TABLES
! NO_ENGINE_SUBSTITUTION
! NO_ZERO_IN_DATE*
! NO_ZERO_DATE*
! ERROR_FOR_DIVISION_BY_ZERO*
! ONLY_FULL_GROUP_BY
@gabidavila

@gabidavila
Example
@gabidavila

@gabidavila
Table `laps`
Column Type
id INT(10)
time (s) INT(10)
summary JSON
speed (m/s) DOUBLE
Gist 🔗
@gabidavila

@gabidavila
SELECT * FROM `laps`;
+----+---------+----------+------+ 
| id | athlete | distance | time | 
+----+---------+----------+------+ 
| 1 | Martha | 3200 | 1800 | 
| 2 | Marcus | 10000 | 2100 | 
| 3 | Martha | 200 | 50 | 
| 4 | James | 1600 | 480 | 
| 5 | Marcus | 1600 | 240 | 
+----+---------+----------+------+ 
@gabidavila

@gabidavila
mysql> SELECT * FROM `laps` GROUP BY athlete;
ERROR 1055 (42000): Expression #1 of SELECT list is not in GROUP BY clause and
contains nonaggregated column 'olympics.laps.id' which is not functionally
dependent on columns in GROUP BY clause; this is incompatible with
sql_mode=only_full_group_by
SELECT * FROM `laps` GROUP BY athlete;
SELECT `id`, `athlete`, `distance`, `time`, `summary`
FROM `laps`
GROUP BY athlete;

@gabidavila
Why?
"Reject queries for which the select list, HAVING condition, or ORDER BY list
refer to nonaggregated columns that are neither named in the GROUP BY
clause nor are functionally dependent on (uniquely determined by) GROUP BY
columns."
@gabidavila

@gabidavila
Translation
This happens because the query is grouping by athlete and MySQL is
magically trying to “discover” which field should it bring as a result to the other
columns since there are more than one possible value per row.
 
This mode is known as ONLY_FULL_GROUP_BY.
More info 🔗
@gabidavila

@gabidavila
Group By
● Columns should either be on the GROUP BY clause OR be using an
aggregator on the SELECT clause.
● Aggregators:
● ANY_VALUE()
● MIN()
● MAX()
● SUM()
● COUNT()

@gabidavila
MySQL `sys` schema
● Needs to be installed on 5.6
● Installed by default on 5.7
● MySQL Workbench 6.3 ships with a client for it
● If enabled, in production should be used for critical analytical cases

@gabidavila
What can I do with it?
*High Cost SQL statements select * from sys.`x$statement_analysis`
Top 5% slower queries
select * from
sys.`x$statements_with_runtimes_in_95th_percentile`
Use temporary tables select * from sys.`statements_with_temp_tables`
Unused Indexes select * from sys.`schema_unused_indexes`
Full table scans
select * from
sys.`schema_tables_with_full_table_scans`
*x$ is actually a view duplicated from a table with a more user friendly format

@gabidavila
Changes on 5.7
● Passwords can now have expiration date
● NO_ZERO_DATE, NO_ZERO_IN_DATE, ERROR_FOR_DIVISION_BY_ZERO
are deprecated and being default in STRICT mode in future versions
● Triggers now supports more than one event per table
● YEAR(2) was deprecated on 5.6, removed on 5.7

@gabidavila
Thank you!
● Twitter: @gabidavila
● Blog: http://gabriela.io
● Freenode: gabidavila
● Feedback: https://joind.in/talk/18d42
● References: MySQL documentation

Diving into MySQL 5.7: advanced features

More Related Content

What's hot (20)

Similar to Diving into MySQL 5.7: advanced features (20)

More from Gabriela Ferrara (14)

Recently uploaded (20)

Diving into MySQL 5.7: advanced features