What's New MySQL 8.0?

Copyright © 2016, Oracle and/or its affiliates. All rights reserved. |
MySQL Server 8.0
Morgan Tocker
MySQL Product Manager (Server)
Copyright © 2017, 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.
3

MySQL 5.7 – Improvements across the board
• Replication
• InnoDB
• Optimizer
• Security
• Performance Schema
• GIS
• Triggers
• Partitioning
• New! SYS Schema
• New! JSON
• Performance
4
200+ new
features
In total!

Copyright © 2017, Oracle and/or its affiliates. All rights reserved. | 5
The MySQL 5.7 Story (cont.)
JSON Type
JSON Functions
Virtual Columns
GIS Improvements
Observability
Manageability

The MySQL 8.0 Story
• Mobile First
• Developer Centric
• Premium placed on time to Market
6
“Making MySQL Better for Modern Applications”

GIS
• Geography Support
• Spatial Reference Systems (SRS) Support
• Standard compliant axis ordering in import/export functions
• Helper functions to manipulate and convert data:
• st_x(geom, x)
• st_y(geom, y)
• st_srid(geom, srid)
7

JSON Functions
8
MySQL 5.7 and 8.0
JSON_ARRAY_APPEND()
JSON_ARRAY_INSERT()
JSON_ARRAY()
JSON_CONTAINS_PATH()
JSON_CONTAINS()
JSON_DEPTH()
JSON_EXTRACT()
JSON_INSERT()
JSON_KEYS()
JSON_LENGTH()
JSON_MERGE_[PATCH]()
JSON_OBJECT()
JSON_QUOTE()
JSON_REMOVE()
JSON_REPLACE()
JSON_SEARCH()
JSON_SET()
JSON_TYPE()
JSON_UNQUOTE()
JSON_VALID()
JSON_PRETTY()
JSON_STORAGE_SIZE()
JSON_STORAGE_FREE() 
JSON_MERGE_PRESERVE()
JSON_ARRAYAGG()
JSON_OBJECTAGG()
JSON_TABLE() *labs

SELECT * FROM seats,
JSON_TABLE(doc, "$.properties.amenities[*]" COLUMNS (
id for ordinality,
amenity_type VARCHAR(100) PATH "$.type",
distance float PATH '$.distance_in_meters')
) AS amenities
WHERE seats.id = 28100
AND amenities.amenity_type IN ('snacks', 'bar')
ORDER BY amenities.distance;
+------+------------------+------------+
| id | amenity_type | distance |
+------+------------------+------------+
| 2 | bar | 100.538 |
| 3 | snacks | 136.647 |
+------+------------------+------------+
2 rows in set (0.00 sec)
9
Labs

utfmb4 as default character set
• Support for the latest Unicode 9.0
• Accent and case sensitive collations
• Language specific collations
• including Japanese and Russian!
10
MySQL 8.0

UTF-8
• UTF-8 is the dominating
character set in today’s
applications
• Requires 1-4 bytes for storing
characters
• Historically a performance
problem
The character set for the Web
11
https://en.wikipedia.org/wiki/UTF-8

MySQL 8.0 vs MySQL 5.7 utf8mb4
12
SELECT DISTINCT_RANGES
0
45000
90000
135000
180000
8 16 64 128 512 1024
OLTP RO
0
10000
20000
30000
40000
8 16 64 128 512 1024
OLTP RW
0
7500
15000
22500
30000
8 16 64 128 512 1024
+300-350% in OLTP RO
+176-233% in OLTP RW
+1500-1800% in SELECT DISTINCT_RANGES

MySQL 8.0 utf8mb4 vs MySQL 5.7 utf8mb3
13
SELECT DISTINCT_RANGES
0
45000
90000
135000
180000
8 16 64 128 512 1024
OLTP RO
0
10000
20000
30000
40000
8 16 64 128 512 1024
OLTP RW
0
7500
15000
22500
30000
8 16 64 128 512 1024
+270-340% in OLTP RO
+150-200% in OLTP RW
+1300-1600% in SELECT DISTINCT_RANGES

New! UUID and Bit-wise Improvements
• Functions to convert UUID to and from binary:
– UUID_TO_BIN()
– BIN_TO_UUID()
– plus IS_UUID()
• Bit-wise operations on binary data types
• Bit-wise operations on binary data types
– Designed with IPv6 in mind:
– INET6_ATON(address) & INET6_ATON(network)
14
Feature Request 
from Developers

UUID_TO_BIN Optimization
15
Insert Performance
26 26.75 27.5 28.25 29
Optimized
Original
• Binary format is now smaller and insert-order efficient:
11e678fe53303f87a4778c89a52c4f3b
53303f87-78fe-11e6-a477-8c89a52c4f3bFrom VARCHAR(36)
To VARBINARY(16)

New! Better Handling of Hot Row Contention
16
SELECT * FROM tickets
WHERE id IN (1,2,3,4)
AND order_id IS NULL
FOR UPDATE
NOWAIT;
SELECT * FROM tickets
WHERE id IN (1,2,3,4)
AND order_id IS NULL
FOR UPDATE
SKIP LOCKED;
Error
immediately if a
row is already
locked
Non
deterministically
skip over locked
rows

New! Common Table Expressions
• “With queries”
• Both Recursive and Non-Recursive Forms
• Simplifies writing complex SQL: 
 
WITH t1 AS (SELECT * FROM tblA WHERE a=‘b’) 
SELECT * FROM t1;
17
Feature Request 
from Developers

Print 1 to 10
Recursive CTE (Example)
a
1
2
3
4
5
6
7
8
9
10
WITH RECURSIVE qn AS 
( SELECT 1 AS a  
UNION ALL 
SELECT 1+a FROM qn WHERE a<10 
) 
SELECT * FROM qn;

New! Window Functions
19
Improving Support for Analytics
Feature Request 
from Developers
SELECT name, department_id, salary,
SUM(salary) 
OVER (PARTITION BY department_id)
AS department_total
FROM employee
ORDER BY department_id, name;

Improved to consider buffer pool fit
Optimizer Cost Model
Feature Request 
from DBAs
SELECT * FROM Country 
WHERE population > 20000000;
Model for a table scan:
# pages in table *
(IO_BLOCK_READ_COST |
MEMORY_BLOCK_READ_COST) 
# records * ROW_EVALUATE_COST
= 25.4 100% in memory
= 29.9 100% on disk
Model for a range scan:
# records_in_range *
(IO_BLOCK_READ_COST |
MEMORY_BLOCK_READ_COST) 
# records_in_range *
ROW_EVALUATE_COST + #
records_in_range *
ROW_EVALUATE_COST  
= 22.5 100% in memory
= 60 100% on disk
Model accounts for
memory ﬁt. For data on disk an IO
block read defaults to 1.0. In
memory defaults to 0.25.
Much larger
performance difference for
range scan not in memory
(good)

New! Histograms
• Track distribution skew of data
• Used for operations in the format: 
COLUMN operator CONSTANT
• Much lower cost than adding an index
Feature Request 
from DBAs

New! Descending Indexes
CREATE TABLE t1 ( 
a INT, 
b INT, 
INDEX a_desc_b_asc (a DESC, b ASC) 
);
22
For B+tree indexes
mysql> EXPLAIN SELECT * FROM t1 ORDER BY a DESC, b ASC;
.. +--------------+---------+------+------+----------+-------------+
.. | key | key_len | ref | rows | filtered | Extra |
.. +--------------+---------+------+------+----------+-------------+
.. | a_desc_b_asc | 10 | NULL | 10 | 100.00 | Using index |
.. +--------------+---------+------+------+----------+-------------+
mysql 8.0> EXPLAIN SELECT * FROM t1 ORDER BY a ASC, b ASC;
.. +--------------+---------+------+------+----------+-----------------------------+
.. | key | key_len | ref | rows | filtered | Extra |
.. +--------------+---------+------+------+----------+-----------------------------+
.. | a_desc_b_asc | 10 | NULL | 10 | 100.00 | Using index; Using filesort |
.. +--------------+---------+------+------+----------+-----------------------------+

New! Invisible Indexes
• Indexes are “hidden” to the MySQL Optimizer
– Not the same as “disabled indexes”
– Contents are fully up to date and maintained by DML
• Two use cases:
– Soft Delete (Recycle Bin)
– Staged Rollout
23
Feature Request 
from DBAs

Example Usage
Soft Delete
• I don’t think this index is used any more: 
ALTER TABLE Country ALTER INDEX c INVISIBLE;
• I need to revert: 
ALTER TABLE Country ALTER INDEX c VISIBLE;
• It is now safe to drop: 
ALTER TABLE Country DROP INDEX c;

Staged Rollout
• Adding any new index can change existing execution plans
• All change introduces risk of regression 
 
ALTER TABLE Country ADD INDEX c (Continent)
INVISIBLE; 
# after some time 
ALTER TABLE Country ALTER INDEX c VISIBLE;
25

SELECT * FROM information_schema.statistics WHERE is_visible='NO'; 
*************************** 1. row *************************** 
TABLE_CATALOG: def 
TABLE_SCHEMA: world 
TABLE_NAME: Country 
NON_UNIQUE: 1 
INDEX_SCHEMA: world 
INDEX_NAME: c 
SEQ_IN_INDEX: 1 
COLUMN_NAME: Continent 
COLLATION: A 
CARDINALITY: 7 
SUB_PART: NULL 
PACKED: NULL 
NULLABLE: 
INDEX_TYPE: BTREE 
COMMENT: disabled 
INDEX_COMMENT: 
IS_VISIBLE: NO

Evolution from MySQL 5.5 to 8.0
Performance Schema
Feature Request 
from DBAs
MySQL 8.0
Histograms
Indexes
Data Locks instrumentation
SQL Errors instrumentation
Variables Instrumentation
Replication
Table plugin
Improved Defaults 
Digest Query Sample
MySQL 5.7
Memory Instrumentation
Prepared Statements Instrumentation
Transactions Instrumentation
Scalable Memory Allocation
Bundled SYS schema
Lower Overhead
MySQL 5.6
Statement Instrumentation
Lower Overhead
MySQL 5.5
Event Waits
Mutexes
Files
Threads

Performance Comparison
Query Time
0 10 20 30 40
MySQL 8.0
MySQL 5.7
Over 30x faster!
SELECT * FROM sys.session 
1000 active sessions
Time in Seconds (Lower is better)

SELECT * FROM test.no_table;
ERROR 1146 (42S02): Table 'test.no_table' doesn't exist
SELECT * FROM performance_schema.events_errors_summary_global_by_error 
WHERE sum_error_handled > 0 OR SUM_ERROR_RAISED > 0G
*************************** 1. row ***************************
ERROR_NUMBER: 1146
ERROR_NAME: ER_NO_SUCH_TABLE
SQL_STATE: 42S02
SUM_ERROR_RAISED: 1
SUM_ERROR_HANDLED: 0
FIRST_SEEN: 2016-09-11 20:52:42
LAST_SEEN: 2016-09-11 20:52:42
1 row in set (0.00 sec)

query: INSERT INTO `t1` VALUES (...)
db: mysqlslap
total_latency: 54.43 s
exec_count: 58377
lock_latency: 1.70 s
..
digest: 4e0c5b796c4052b0da4548fd7cb694be
first_seen: 2017-04-16 20:59:16
last_seen: 2017-04-16 21:00:34
latency_distribution:
0us+
10us+ #############################
100us+ ####################
1ms+ #
10ms+
100ms+
1s+
10s+
New! Performance Schema Histograms
30
+---------+--------------------------------+-------+
| bucket | visualization | count |
+---------+--------------------------------+-------+
| 0us+ | # | 1253 |
| 10us+ | ############################## | 43102 |
| 100us+ | ################# | 25013 |
| 1ms+ | # | 2003 |
| 10ms+ | | 325 |
| 100ms+ | | 17 |
| 1s+ | | 0 |
| 10s+ | | 0 |
+---------+--------------------------------+-------+
Showing distribution of query time from a run of mysqlslap
Generated with a quick CTE over
events_statements_histogram_global
Feature Request 
from DBAs
Available on a per
statement digest level. Can quickly
aggregate top-N statements with
latency distribution.

New! Persist Configuration
• Persist GLOBAL Server Variables
– SET PERSIST max_connections = 500;
– SET PERSIST_ONLY innodb_log_file_size = 128*1024*1024;
• Examples Include:
– Offline_mode
– Read_Only
• Requires no filesystem access
Cloud Friendly

New! Variables Info
33
Find the source of variables changed on your installation
SELECT * FROM performance_schema.variables_info WHERE variable_source != 'COMPILED';
+------------------------+-----------------+--------------------+-----------+-----------+---------------------+----------+-----------+
| VARIABLE_NAME | VARIABLE_SOURCE | VARIABLE_PATH | MIN_VALUE | MAX_VALUE | SET_TIME | SET_USER | SET_HOST |
+------------------------+-----------------+--------------------+-----------+-----------+---------------------+----------+-----------+
| autocommit | DYNAMIC | | 0 | 0 | 2017-04-16 20:56:53 | msandbox | localhost |
| basedir | COMMAND_LINE | | 0 | 0 | 2017-04-16 21:08:11 | | |
| bind_address | EXPLICIT | [..]my.sandbox.cnf | 0 | 0 | 2017-04-16 21:08:11 | | |
| character_set_client | DYNAMIC | | 0 | 0 | 2017-04-16 20:56:53 | msandbox | localhost |
| character_set_results | DYNAMIC | | 0 | 0 | 2017-04-16 20:56:53 | msandbox | localhost |
| collation_connection | DYNAMIC | | 0 | 0 | 2017-04-16 20:56:53 | msandbox | localhost |
| datadir | COMMAND_LINE | | 0 | 0 | 2017-04-16 21:08:11 | | |
| foreign_key_checks | DYNAMIC | | 0 | 0 | 2017-04-16 20:56:53 | msandbox | localhost |
| log_error | COMMAND_LINE | | 0 | 0 | 2017-04-16 21:08:11 | | |
| lower_case_table_names | EXPLICIT | [..]my.sandbox.cnf | 0 | 2 | 2017-04-16 21:08:11 | | |
| pid_file | COMMAND_LINE | | 0 | 0 | 2017-04-16 21:08:11 | | |
| plugin_dir | COMMAND_LINE | | 0 | 0 | 2017-04-16 21:08:11 | | |
| port | COMMAND_LINE | | 0 | 65535 | 2017-04-16 21:08:11 | | |
| socket | COMMAND_LINE | | 0 | 0 | 2017-04-16 21:08:11 | | |
| sql_mode | DYNAMIC | | 0 | 0 | 2017-04-16 20:56:53 | msandbox | localhost |
| sql_notes | DYNAMIC | | 0 | 0 | 2017-04-16 20:56:53 | msandbox | localhost |
| time_zone | DYNAMIC | | 0 | 0 | 2017-04-16 20:56:53 | msandbox | localhost |
| tmpdir | EXPLICIT | [..]my.sandbox.cnf | 0 | 0 | 2017-04-16 21:08:11 | | |
| unique_checks | DYNAMIC | | 0 | 0 | 2017-04-16 20:56:53 | msandbox | localhost |
+------------------------+-----------------+--------------------+-----------+-----------+---------------------+----------+-----------+

New! Transactional Data Dictionary
• Crash-safe Database
• Common data dictionary for
the server and InnoDB
• Crash-safe & Atomic DDL
• CREATE USER <userlist>, DROP
DATABASE with all-or-none
semantic
• Simplifies replication failure
cases
34
• Meta-data locking for FK
constraints
• FK moved from InnoDB to
server layer
• Scalable Information Schema
• Queries are now executed as
set of SQL views on tables
• Large performance
improvements

30x Faster
SELECT TABLE_SCHEMA, 
TABLE_NAME, TABLE_TYPE, 
ENGINE, ROW_FORMAT 
FROM information_schema.tables 
WHERE TABLE_SCHEMA LIKE 'db%';
35
Test Performed with 100 schemas, each with 50 tables.

Defaults
• Event Scheduler On
• Utf8mb4 Character Set
• Performance Schema: Memory,
Transactions, MDL
• InnoDB: autoinc_lock_mode=2,
flush_neighbors=0,
max_dirty_pages_pct_lwm=10,
max_dirty_pages_pct
default=90
• Improvements to back_log
auto-calculation
• max_error_count=1024
• max_allowed_packet=64M
36
Server-Core

Defaults (cont.)
• Binary Log, log-slave-updates on
• {master,relay-log}—info-repository = TABLE
• transaction-write-set-extraction = XXHASH64
• slave-rows-search-algorithms = ‘INDEX_SCAN,HASH_SCAN’
• expire_logs_days=30
• server-id=1
37
Replication

InnoDB Dedicated Server
• Great for Virtual Machines & Cloud Deployments
• Auto scales Log File Size, Buffer Pool Size, Flush Method
• Easiest way to enable: 
 
SET PERSIST_ONLY innodb_dedicated_server = TRUE
38

Performance Improvements
• Improved Query Consistency
• Histograms
• Improved Cost Model
• Faster Table/Range Scans
• FORCE INDEX skips index
dives where possible
• Expanded Hints Support
• JSON Partial Update
(Optimizer, Replication)
39
• Parallel Replication
• UTF8MB4
• Information Schema
• Performance Schema Indexes

QueriesperSecond
0
40,000
80,000
120,000
160,000
Users
1 2 4 8 16 32 64 128 256 512
MySQL 8.0 Labs
MySQL 5.7
MySQL 5.6
MySQL 8.0 Labs: Sysbench Benchmark: IO-bound OLTP RW Updates-only 
Intel(R) Xeon(R) Platinum 8168 CPU
2CPU-sockets, 48cores-HT
2x Intel Optane
Oracle Linux 7.3

QueriesperSecond
0
300,000
600,000
900,000
1,200,000
Users
1 2 4 8 16 32 64 128 256 512
MySQL 8.0 Labs
MySQL 5.7
MySQL 5.6
MySQL 8.0 Labs: Sysbench Benchmark: IO-bound OLTP RO Point-Selects 
Intel(R) Xeon(R) Platinum 8168 CPU
2CPU-sockets, 48cores-HT
2x Intel Optane
Oracle Linux 7.3
1,000,000+
QPS

MySQL 8.0 RC: Resource Group Example
QueriesperSecond
0
40,000
80,000
120,000
160,000
No Resource Group With Resource Group
Select
Update
System Configuration :
Oracle Linux 7,
Intel(R) Xeon(R) CPU E7-4860 2.27GHz
40 cores-HT
(40 Cores Shared) (40 Cores for Select)
(10 Cores for Update RG)

• Source code now documented with Doxygen
• New Plugin Infrastructure
• Improved BLOB Storage
• Improved Memcached Interface
• InnoDB Auto Increment Persists
• Parser Refactoring
• Improvements to Temporary Tables
• C++11 and Toolchain Improvements
• GTID_PURGED always settable
• Undo tablespaces now dynamic
• In memory storage engine
• SQL Roles Support
• Encryption for Redo and General Tablespaces
• InnoDB Cats lock scheduler algorithm
• Improved Parallel Replication
• SQL Grouping Function
• Optimizer Trace detailed sort statistics
• Smaller Package Downloads
• Improved usability of cost constant configuration
43
All these features plus…

What's New MySQL 8.0?

Recommended

More Related Content

What's hot (20)

Similar to What's New MySQL 8.0? (20)

More from OracleMySQL (9)

Recently uploaded (20)

What's New MySQL 8.0?