The document provides an overview of the InnoDB storage engine used in MySQL. It discusses InnoDB's architecture including the buffer pool, log files, and indexing structure using B-trees. The buffer pool acts as an in-memory cache for table data and indexes. Log files are used to support ACID transactions and enable crash recovery. InnoDB uses B-trees to store both data and indexes, with rows of variable length stored within pages.

1. <Insert Picture Here>
The InnoDB Storage Engine for MySQL
Morgan Tocker, MySQL Community Manager
http://www.tocker.ca/

2. 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. 4 Years of MySQL Innovation
MySQL Cluster 7.3
MySQL Workbench 6.0
MySQL Migration Wizard
<Insert Picture Here>
MySQL 5.5
MySQL 5.6
Windows installer & Tools MySQL 5.7 M y S Q L
Cluster
MySQL
MySQL Enterprise Monitor 2.3 & e r
M a n a g 3.0
Applier for
Hadoop
MySQL Enterprise Backup
Security
MySQL Utilities
MySQL Workbench 5.2 & 6.0
Scalability
MySQL Cluster 7.2
HA
MySQL Enterprise
MySQL Cluster 7.1
O r a c l e C e r t i f i c a t i o n s Audit

4. Hello and Welcome!
• I will be talking about InnoDB’s internal behaviour.
• Not talking (much) about MySQL.
• Aim of this talk is to give you X-ray vision.
• i.e. not so many direct takeaways, but one day it
will help you debug a problem.

5. Prerequisites
Copyright
©
2012
Oracle
and/or
its
affiliates.
All
rights
reserved.

6. MySQL Architecture

7. IO Performance
L1 cache reference
Branch mispredict
L2 cache reference
Mutex lock/unlock
Main memory reference
Compress 1K bytes with Zippy
Send 2K bytes over 1 Gbps network
Read 1 MB sequentially from memory
Round trip within same datacenter
Disk seek
Read 1 MB sequentially from disk
Send packet CA->Netherlands->CA
0.5 ns!
5 ns!
7 ns!
25 ns!
100 ns!
3,000 ns!
20,000 ns!
250,000 ns!
500,000 ns!
10,000,000 ns!
20,000,000 ns!
150,000,000 ns
See: http://www.linux-mag.com/cache/7589/1.html and Google http://
www.cs.cornell.edu/projects/ladis2009/talks/dean-keynote-ladis2009.pdf

8. IO Performance (cont.)
• 5-10ms per disk IO.
• Maybe 50us for a high end SSD.
• Still not “memory speed”.

9. Buffered IO
• Operating Systems compensate well already.
• Reads are cached with free memory.
• Writes don’t happen instantly.
• A step is introduced to rewrite and merge.
Block 9, 10, 1, 4, 200, 5.
Block 1, 4, 5, 9, 10, 200

10. fsync
Synopsis
#include <unistd.h>
int fsync(int fd);
int fdatasync(int fd);
!
Description
fsync() transfers ("flushes") all modified in-core data of (i.e., modified buffer cache pages for) the file referred to by
the file descriptor fd to the disk device (or other permanent storage device) where that file resides. The call blocks
until the device reports that the transfer has completed. It also flushes metadata information associated with the file
(see stat(2)).

11. Basic
Operation
Copyright
©
2012
Oracle
and/or
its
affiliates.
All
rights
reserved.

12. Buffer Pool
Data Dictionary Cache
Buffer Pool Flush List
Adaptive Hash Indexes
Log Group
ibdata1
space 0
Storage
Buffer Pool LRU
Page Cache
Additional Mem Pool
Log Buffer
Transaction
System
Caching
InnoDB High Level Overview
IBUF_HEADER
IBUF_TREE
TRX_SYS
FIRST_RSEG
DICT_HDR
iblogfile0
iblogfile1
Doublewrite Buffer
Data Dict.
Block 1 (64 pages)
Block 2 (64 pages)
SYS_TABLES
SYS_COLUMNS
SYS_INDEXES
SYS_FIELDS
iblogfile2
Tables with
file_per_table
A.ibd
B.ibd
C.ibd
In Memory
On Disk

13. Query (pages not in buffer pool)
mysqld
SELECT * FROM a
WHERE id = 10;
Buffer Pool
Buffer Pool LRU
Adaptive Hash Indexes
Data Dictionary Cache
Buffer Pool Flush List
Page Cache
Log Group
ibdata1
space 0
Storage
InnoDB
Additional Mem Pool
Log Buffer
Transaction
System
Caching
Not Found
IBUF_HEADER
IBUF_TREE
TRX_SYS
FIRST_RSEG
DICT_HDR
iblogfile0
iblogfile1
Doublewrite Buffer
Data Dict.
Block 1 (64 pages)
Block 2 (64 pages)
SYS_TABLES
SYS_COLUMNS
SYS_INDEXES
SYS_FIELDS
iblogfile2
Tables with
file_per_table
A.ibd
B.ibd
C.ibd

14. Query (pages in buffer pool)
mysqld
SELECT * FROM a
WHERE id = 10;
Buffer Pool
Data Dictionary Cache
Buffer Pool Flush List
Adaptive Hash Indexes
Log Group
ibdata1
space 0
Storage
Buffer Pool LRU
Page Cache
Additional Mem Pool
Log Buffer
Transaction
System
Caching
InnoDB
IBUF_HEADER
IBUF_TREE
TRX_SYS
FIRST_RSEG
DICT_HDR
iblogfile0
iblogfile1
Doublewrite Buffer
Data Dict.
Block 1 (64 pages)
Block 2 (64 pages)
SYS_TABLES
SYS_COLUMNS
SYS_INDEXES
SYS_FIELDS
iblogfile2
Tables with
file_per_table
A.ibd
B.ibd
C.ibd

15. Update Query in a Transaction (simplified)
mysqld
UPDATE a SET col1 =
‘new’ WHERE id = 10;
commit;
Buffer Pool
Data Dictionary Cache
Buffer Pool Flush List
Adaptive Hash Indexes
Log Group
ibdata1
space 0
Storage
Buffer Pool LRU
Page Cache
Additional Mem Pool
Log Buffer
Transaction
System
Caching
InnoDB
IBUF_HEADER
IBUF_TREE
TRX_SYS
FIRST_RSEG
DICT_HDR
iblogfile0
iblogfile1
Doublewrite Buffer
Data Dict.
Block 1 (64 pages)
Block 2 (64 pages)
SYS_TABLES
SYS_COLUMNS
SYS_INDEXES
SYS_FIELDS
iblogfile2
Tables with
file_per_table
A.ibd
B.ibd
C.ibd

16. Log files
• Provide recovery.
• Only written to in regular operation.
• Read only required if there is a crash.
• Are rewritten over-and-over again.
• Think of it like a tank tread.

17. Log files (cont.)
• Are an optimization!
• 512B aligned sequential writes.
• Tablespace writes are 16KiB random writes.
• Tablespace writes to same pages in close time
window can be merged.
• Just need a large enough log file.

18. Checkpoint (Background Activity)
mysqld
(nothing)
Buffer Pool
Data Dictionary Cache
Buffer Pool Flush List
Adaptive Hash Indexes
Log Group
ibdata1
space 0
Storage
Buffer Pool LRU
Page Cache
Additional Mem Pool
Log Buffer
Transaction
System
Caching
InnoDB
IBUF_HEADER
IBUF_TREE
TRX_SYS
FIRST_RSEG
DICT_HDR
iblogfile0
iblogfile1
Doublewrite Buffer
Data Dict.
Block 1 (64 pages)
Block 2 (64 pages)
SYS_TABLES
SYS_COLUMNS
SYS_INDEXES
SYS_FIELDS
iblogfile2
Tables with
file_per_table
A.ibd
B.ibd
C.ibd

19. FAQ
• Q: What do we write to the log - is it committed data
only, or can we write uncommitted data as well?
• A: Both.

20. FAQ
• Q: How do you unapply transactions?
• A: UNDO space.
Think of it like a hidden table internally stored in
ibdata1.

21. Update Query (More Accurate*)
mysqld
UPDATE a SET col1 =
‘new’ WHERE id = 10;
commit;
Page Cache
Buffer Pool
Data Dictionary Cache
Buffer Pool Flush List
Adaptive Hash Indexes
Additional Mem Pool
Redirect older version
Update any indexes
Modify row in place
of row to undo space
to hold both versions.
Log Group
ibdata1
space 0
Storage
Buffer Pool LRU
Page Cache
Log Buffer
Transaction
System
Caching
InnoDB
IBUF_HEADER
IBUF_TREE
TRX_SYS
FIRST_RSEG
DICT_HDR
iblogfile0
iblogfile1
Doublewrite Buffer
Data Dict.
Block 1 (64 pages)
Block 2 (64 pages)
SYS_TABLES
SYS_COLUMNS
SYS_INDEXES
SYS_FIELDS
iblogfile2
Tables with
file_per_table
A.ibd
B.ibd
C.ibd

22. Update Query (cont.)
• Background purge process is able to clean old rows
from UNDO as soon as oldest transaction advances
forward.

23. Summarized Performance
Characteristics
• Log Files:
• Are short sequential writes.
• They permit InnoDB to delay tablespace writes enabling more merging/optimization.
• Buffer Pool:
• “In memory version of the tablespace”.
• Loading/unloading via modified LRU algorithm.

24. Index Structure
• Indexes and “data” in InnoDB are B+Trees.
• Clustered Index design means that data itself is
stored in an index.

25. Index Structure (cont.)
Empty root
Infimum
Level 0
Root
Page 3
Next
Record
Supremum

26. Index Structure (cont.)
Insert: 1
Infimum
Level 0
Root
Page size is 16KB
B* is 2KB
Page 3
1
B*
Supremum

27. Index Structure (cont.)
Insert: 1 to 7
I
Level 0
Root
S
Page 3
1
B*
2
B*
3
B*
4
B*
5
B*
6
B*
7
B*

28. Index Structure (cont.)
Insert: 8
Infimum
Page 3
Level 1
Root
1
4
I
Level 0
Leaf
Allocate new page and link in root
Move records to new page
Split new page
Supremum
S
Page 4
1
B*
2
B*
3
B*
4
B*
5
B*
6
B*
7
B*

29. Index Structure (cont.)
Insert: 8 (Cont.)
Infimum
Level 1
Root
1
4
I
Level 0
Leaf
S
Page 4
1
B*
Split at the middle of original page
2
B*
3
B*
Supremum
Page 3
4
5
I
S
Page 5
4
B*
5
B*
6
B*
7
B*

30. Index Structure (cont.)
Insert: 9 and 10
Infimum
Level 1
Root
1
4
I
Level 0
Leaf
S
Page 4
1
B*
2
B*
3
B*
Supremum
Page 3
4
5
I
S
Page 5
4
B*
5
B*
6
B*
7
B*
8
B*
9
B*
10
B*

31. Index Structure (cont.)
Insert: 11
Infimum
Level 1
Root
4
5
1
4
I
Level 0
Leaf
Page 3
S
Page 4
I
Supremum
11
6
S
Page 5
I
S
Page 6
1
B*
2
B*
3
B*
Insert leads to a split at the insertion point
4
B*
5
B*
6
B*
7
B*
8
B*
9
B*
10
B*
11
B*

32. Index Structure
Level 2
Root
Infimum
Level 1
Internal
Page 6
0
A
1
B
S
Page 4
≥0
→
6
I
Level 0
Leaf
Next
Record
≥0
→
4
I
S
Supremum
Page 3
≥4
→
5
I
Next Page
≥2
→
7
≥4
→
8
Prev Page
I
Page 7
2
C
3
D
S
I
Page 8
4
E
5
F
S
Page 5
S
≥6
→
9
I
Page 9
6
G
7
H
S

33. Page Format
0
38
FIL Header (38)
Other headers and page data,
depending on page type.
Total usable space: 16,338 bytes.
16376
16384
FIL Trailer (8)

34. Row Format
N-5
N-4
N-2
N
N+k
N+k+6
N+k+13
N+k+13+j
Variable field lengths (1-2 bytes per var. field)
Info Flags (4 bits)
Number of Records Owned (4 bits)
Order (13 bits)
Record Type (3 bits)
Next Record Offset (2)
Cluster Key Fields (k)
Transaction ID (6)
Roll Pointer (7)
Non-Key Fields (j)

35. Conclusion
• Page is basic unit of storage.
• Default is 16KiB
• Rows of variable length.

36. Two more useful features
• Adaptive hash - Partial hash index to accelerate
secondary key lookups.
• Change buffering - when non-unique indexes are
not in memory, changes can be temporarily buffered
until they are.

37. Query (by secondary key)
mysqld
SELECT * FROM a
WHERE b_key = 10;
Buffer Pool
Data Dictionary Cache
Buffer Pool Flush List
Adaptive Hash Indexes
Log Group
ibdata1
space 0
Storage
Buffer Pool LRU
Page Cache
Additional Mem Pool
Log Buffer
Transaction
System
Caching
InnoDB
IBUF_HEADER
IBUF_TREE
TRX_SYS
FIRST_RSEG
DICT_HDR
iblogfile0
iblogfile1
Doublewrite Buffer
Data Dict.
Block 1 (64 pages)
Block 2 (64 pages)
SYS_TABLES
SYS_COLUMNS
SYS_INDEXES
SYS_FIELDS
iblogfile2
Tables with
file_per_table
A.ibd
B.ibd
C.ibd

38. Update Query (large table)
mysqld
UPDATE a SET col1 =
‘new’ WHERE id = 10;
commit;
Buffer Pool
Caching
InnoDB
Buffer Pool LRU
Adaptive Hash Indexes
Data Dictionary Cache
Buffer Pool Flush List
Page Cache
Additional Mem Pool
Not Required
Log Group
ibdata1
space 0
Storage
Transaction
System
Log Buffer
IBUF_HEADER
IBUF_TREE
TRX_SYS
FIRST_RSEG
DICT_HDR
iblogfile0
iblogfile1
Doublewrite Buffer
Data Dict.
Block 1 (64 pages)
Block 2 (64 pages)
SYS_TABLES
SYS_COLUMNS
SYS_INDEXES
SYS_FIELDS
iblogfile2
Tables with
file_per_table
A.ibd
B.ibd
C.ibd

39. New
Features
Copyright
©
2012
Oracle
and/or
its
affiliates.
All
rights
reserved.

40. MySQL 5.5+
•
•
•
•
•
•
•
•
•
IO Scalability
Async IO
Multiple Buffer Pools
Adaptive Flushing
Scan Resistant LRU
Compressed Pages
CPU Scalability
Improved Atomics
Spin Loops with PAUSE

41. MySQL 5.6+
• LRU Dump and
Restore
• Improved Group
Commit
• Fulltext Search
• Fast Read-only
Transactions
• Memcached Interface
• Information Schema
metadata tables
•
•
•
•
Persistent Statistics
Variable Page Size
Online DDL
Transportable
Tablespace
• Transactional
Replication
Using InnoDB

42. MySQL 5.7+
• Faster Temporary Tables
• Index Lock Contention Reduction
• More Online DDL
• Extend VARCHAR
• Rename Index
• Improved Read-Only Transactions
• Improved CPU Scalability

43. Configuration
Copyright
©
2012
Oracle
and/or
its
affiliates.
All
rights
reserved.

44. The Top 3
1. innodb-buffer-pool-size
2. innodb-log-file-size
3. innodb_flush_log_at_trx_commit

45. innodb-buffer-pool-size
• Really only one major buffer/cache settings to set.
• Responsible for all pages types (data, indexes, undo,
insert buffer..)

46. innodb-buffer-pool-size (cont.)
• Recommendation is 50-80% of RAM.
• Default is 128M of RAM.
• Please allow 5-10% on top for other meta data to
grow.

47. innodb-log-file-size
• Log files are on disk, but this contributes to how
many unflushed (dirty) pages you can hold in
memory.
• In theory larger log files = longer crash recovery.
• In MySQL 5.5 -2G max.
• In MySQL 5.6 - 4G is usually safe.
• Early versions should be much smaller.
• Default is 48M Log Files.

48. innodb_flush_log_at_trx_commit
• Default is full ACID Compliance (=1)
• Can be set to 0/2 if you do not mind some data loss.

49. innodb_flush_log_at_trx_commit
• 0 = Log buffer written + synced once per second.
Nothing done at commit.
• 1 = Log buffer written + synced once per second +
written and synced on commit.
• 2 = Log buffer written + synced once per second +
written (not synced) on commit.
!
2 is a slightly safer version of 0.

50. Requires about 5-10% of buffer pool
size as overhead (not directly
configurable).
Basic Configuration
Buffer Pool
InnoDB
Buffer Pool LRU
Adaptive Hash Indexes
Additional Mem Pool
innodb_log_file_size. Typical
values 256M+. Default of 2 files
(innodb_log_files_in_group).
Log Group
iblogfile0
ibdata1
space 0
innodb_log_buffer_size. Typical
Doublewrite Buffer
values 1-8M. Flushing
IBUF_HEADER
IBUF_TREE
Block 1 (64 pages)
behaviour influenced by
TRX_SYS
Block 2 (64 pages)
FIRST_RSEG
innodb_flush_log_at_trx_commit.
DICT_HDR
Storage
Data Dictionary Cache
Buffer Pool Flush List
Page Cache
Log Buffer
Transaction
System
Caching
Innodb_buffer_pool_size.
Recommendation is 50-80% RAM.
iblogfile1
iblogfile2
Tables with
file_per_table
Data Dict.
SYS_TABLES
SYS_COLUMNS
SYS_INDEXES
SYS_FIELDS
A.ibd
B.ibd
C.ibd
innodb_file_per_table
(Default: ON in 5.6+)

51. The ~Top 10
1. innodb-buffer-pool-size
2. innodb-log-file-size
3. innodb-log-buffer-size
4. innodb_flush_log_at_trx_commit
5. innodb_flush_method
6. innodb_flush_neighbors
7. innodb_io_capacity, innodb_io_capacity_max,
innodb_lru_scan_depth
8. innodb-buffer-pool-instances
9. innodb_read_io_threads and innodb_write_io_threads

52. Less-likely to need configuration
•
•
•
•
•
innodb_thread_concurrency
innodb_concurrency_tickets
innodb_max_pct_dirty_pages
innodb_use_native_aio (always on)
innodb_old_blocks_time (5.6 default: 1000)

53. Deprecated Settings
• Typically “remove on sight” from config files:
• innodb_additional_mempool_size
• innodb_use_sys_malloc

54. Credits
• InnoDB Architecture Diagrams via https://github.com/
jeremycole/innodb_diagrams
• Available under (3-clause) BSD license
Copyright (c) 2013, Twitter, Inc.
Copyright (c) 2013, Jeremy Cole <jeremy@jcole.us>
Copyright (c) 2013, Davi Arnaut
<darnaut@gmail.com>