Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.
Session 101100: Database Compression for JD Edwards
Paul Shearer, Velocity, JD Edwards Solution Engineer
©2013 Velocity Te...
Agenda
• What is DB compression?
• How does DB compression work?
• What are the benefits of DB compression for EnterpriseO...
What is DB Compression?
DB compression is like converting English text into Chinese.
压缩就像是将英文文本转换成中国人
English = 59 charact...
Oracle Compression

EXPERIENCE. INNOVATION.

Velocity Proprietary and Confidential

4
SQL Compression
• Column Prefix Compression
– Find repeating ‘byte pattern’ in the same column in the same
data page

• Di...
Column Prefix Compression

EXPERIENCE. INNOVATION.

Velocity Proprietary and Confidential

6
Column Prefix Compression

EXPERIENCE. INNOVATION.

Velocity Proprietary and Confidential

7
Benefits of DB Compression for E1
• Disk space reduction
– 60% to 70% for Oracle DBs
– 70% to 90% reduction for SQL DBs

•...
DB Requirements for Compression

• SQL Enterprise Edition
– 2008 and higher

• Oracle Enterprise Edition + Feature
– 11G a...
JDE Test Environment
•

Server
– OracleVM 2.1.2
– Oracle Enterprise Linux 5 64 bits Paravirtualized
– 4 cores & 16 GB of m...
JDE Test - Implementation
•

Created a separated compressed tablespace

•

Moved top tables to new tablespace

•

Utilized...
Test Results – Disk Space
Storage Utilization

~60% Storage Savings

1200
1000
800
MB

600
400
200
0
Uncompressed

EXPERIE...
Test Results – Read Performance
Read Performance - Full Scans
Compression is 2.7x
Faster

800
700
600
Time
500
(seconds)
4...
Test Results – Write Performance
DML Performance - Inserts
Compression has ~1%
Overhead

23000

22000
Time
(seconds)
21000...
Test Results – Write Performance
DML Performance - Updates

Compression has ~2%
Overhead

23000

22000
Time
(seconds)
2100...
Sample Compression Rates
Table

Name

Compression

F4111

Item Ledger File

76%

F0911

Account Ledger

73%

F0411

Accoun...
SQL Compression Lab Project Results

EXPERIENCE. INNOVATION.

Velocity Proprietary and Confidential

17
Environment
•

•

•

•

IBM 3850 M3
• 128gb RAM
• 16 Cores
SQL 2008 Enterprise x64 SP1
• 47.5gb memory allocated
• Paralle...
Implementation
•

•
•

Restored JDE_PRODUCTION to SQL2008POC on SQL21 and created three
databases
• Uncompressed
• Row Com...
Database Size
Database Sizes
180
160

162

140
Size in GB

120

~90% Reduction in
Storage Needed

100
80

64

60
40
16

20...
Database Backup Size
Backup Sizes
18
16

15.3

14

Size in GB

12
10
8

6.9

6

5

~67% Reduction in
Backup Storage
Needed...
Backup Times
Backup Times
30
25

24

Minutes

20

~93% Reduction in
Backup Times

15
10
5.5
5

1.5

0
Uncompressed

Row Co...
Restore Times
Restore Times
30
25

24

Minutes

20

~81% Reduction in
Restore Times

15
10
5.5
5

1.5

0
Uncompressed

Row...
Number of Reads
Number of Reads
50000
45000

45671

40000
35000
Reads

30000

~76% Reduction in
Number of Reads

25000
200...
Read Time
Read Time (Seconds)
4500
4000

3995

3500

Seconds

3000
2500

~93% Reduction in
Read Time

2000
1500
1000

502
...
Number of Writes
Number of Writes
21000

20792

20500

20272

20000

Writes

19500
19000
18386

18500

~11% Reduction in
N...
Write Time
Write Time (Seconds)
250

220

200

Seconds

165

154

150

~24% Reduction in
Write Time

100
50
0
Uncompressed...
CPU Utilization
Percent CPU Time
17

16.80666781
16.38710853

Percent

16.5

16

15.5

~8% Increase in
Percent CPU Time
Us...
Memory Utilization
Memory Utilization
2500000
2164773.926

MB Allocated

2000000
1500000

~73% Reduction in
Memory Needed
...
Buffer Cache Hit Ratio
Buffer Cache Hit Ratio
98

97.40602574

97.5
97

96.52317432

96.5
Percent

96

~3% Increase in
Buf...
Disk Utilization
Avg. Disk sec/Transfer
0.025
0.021782055

Percent

0.02
0.014864164

0.015

0.015834584

~33% Decrease in...
Top 5 LRQ - Duration
Top 5 LRQ – Duration (Seconds)
200
180

180.92578

160

Seconds

140
120
100

86.04883
71.97558

80

...
Top 5 LRQ – CPU Time (ms)
Top 5 LRQ - CPU Time (Seconds)
60.00
50.00

49.43
40.44629

Seconds

40.00

33.87402

~31% Reduc...
Top 5 LRQ – Physical Reads
Top 5 LRQ - Physical Reads
25000
20588

Physical Reads

20000

15000

~77% Reduction in
Physica...
Top 5 LRQ - Writes
Top 5 LRQ - Writes
6000
5302
5000

4617
4204

Writes

4000

~9% Reduction in
Writes

3000
2000
1000
0
U...
Notes
•

Recommended Indexes
• Index’s are used more efficiently when the DB is compressed
• Uncompressed – 40 new indexes...
International Footprint to Serve Our Customers

Glasgow, Scotland
Seattle, Washington
Minneapolis, Minnesota
Tukwila, Wash...
Velocity Is an Active Member of the
Oracle® Community
Built first
hosted JD
Edwards
solution
Support for
Oracle edge
apps ...
Questions?

EXPERIENCE. INNOVATION.

Velocity Proprietary and Confidential

39
Thank you.
For more information:
Paul.Shearer@Velocity.cc
803-386-9680
©2013 Velocity Technology Solutions, Inc.

EXPERIEN...
Upcoming SlideShare
Loading in …5
×

EnterpriseOne and Database Compression

913 views

Published on

Oracle EnterpriseOne performance can be substantially improved by implementing database compression on Oracle or SQL Server databases. This presentation outlines best practices.

Published in: Education, Technology
  • Be the first to comment

  • Be the first to like this

EnterpriseOne and Database Compression

  1. 1. Session 101100: Database Compression for JD Edwards Paul Shearer, Velocity, JD Edwards Solution Engineer ©2013 Velocity Technology Solutions, Inc. EXPERIENCE. INNOVATION. EXPERIENCE. INNOVATION. Velocity Proprietary and Confidential 1
  2. 2. Agenda • What is DB compression? • How does DB compression work? • What are the benefits of DB compression for EnterpriseOne? • Which database editions support compression? • Case Studies! EXPERIENCE. INNOVATION. Velocity Proprietary and Confidential 2
  3. 3. What is DB Compression? DB compression is like converting English text into Chinese. 压缩就像是将英文文本转换成中国人 English = 59 characters Chinese = 16 characters “Compression uses symbols to replace long, reoccurring patterns.” EXPERIENCE. INNOVATION. Velocity Proprietary and Confidential 3
  4. 4. Oracle Compression EXPERIENCE. INNOVATION. Velocity Proprietary and Confidential 4
  5. 5. SQL Compression • Column Prefix Compression – Find repeating ‘byte pattern’ in the same column in the same data page • Dictionary Page Compression – Find repeating ‘byte patterns’ across columns within the same data page EXPERIENCE. INNOVATION. Velocity Proprietary and Confidential 5
  6. 6. Column Prefix Compression EXPERIENCE. INNOVATION. Velocity Proprietary and Confidential 6
  7. 7. Column Prefix Compression EXPERIENCE. INNOVATION. Velocity Proprietary and Confidential 7
  8. 8. Benefits of DB Compression for E1 • Disk space reduction – 60% to 70% for Oracle DBs – 70% to 90% reduction for SQL DBs • Lower RAM requirements for the DBMS* – Data pages are cached in RAM – Increase Buffer-hit-cache ratio – Fewer physical disk IOs • 50+% performance gains on EnterpriseOne performance – Each physical read operation returns more data EXPERIENCE. INNOVATION. Velocity Proprietary and Confidential 8
  9. 9. DB Requirements for Compression • SQL Enterprise Edition – 2008 and higher • Oracle Enterprise Edition + Feature – 11G and higher EXPERIENCE. INNOVATION. Velocity Proprietary and Confidential 9
  10. 10. JDE Test Environment • Server – OracleVM 2.1.2 – Oracle Enterprise Linux 5 64 bits Paravirtualized – 4 cores & 16 GB of memory • Oracle Database – Oracle DB 11g Enterprise Edition (11.1.0.7) – Block Size: 8K (default) – Memory Target (SGA + PGA): 12 GB EXPERIENCE. INNOVATION. Velocity Proprietary and Confidential 10
  11. 11. JDE Test - Implementation • Created a separated compressed tablespace • Moved top tables to new tablespace • Utilized compress for all operations • Leveraged online redefinition to move data • Compared results before and after compression EXPERIENCE. INNOVATION. Velocity Proprietary and Confidential 11
  12. 12. Test Results – Disk Space Storage Utilization ~60% Storage Savings 1200 1000 800 MB 600 400 200 0 Uncompressed EXPERIENCE. INNOVATION. Compressed Velocity Proprietary and Confidential 12
  13. 13. Test Results – Read Performance Read Performance - Full Scans Compression is 2.7x Faster 800 700 600 Time 500 (seconds) 400 300 200 Uncompressed EXPERIENCE. INNOVATION. Compressed Velocity Proprietary and Confidential 13
  14. 14. Test Results – Write Performance DML Performance - Inserts Compression has ~1% Overhead 23000 22000 Time (seconds) 21000 20000 Uncompressed EXPERIENCE. INNOVATION. Compressed Velocity Proprietary and Confidential 14
  15. 15. Test Results – Write Performance DML Performance - Updates Compression has ~2% Overhead 23000 22000 Time (seconds) 21000 20000 Uncompressed EXPERIENCE. INNOVATION. Compressed Velocity Proprietary and Confidential 15
  16. 16. Sample Compression Rates Table Name Compression F4111 Item Ledger File 76% F0911 Account Ledger 73% F0411 Accounts Payable Ledger 69% F43121 Purchase Order Receiver File 69% F4311 Purchase Order Detail File 68% F9000 Task Master 62% F0101 Address Book Master 61% F4229 Sales Summary History File 56% F0005 User Defined Code Types 54% EXPERIENCE. INNOVATION. Velocity Proprietary and Confidential 16
  17. 17. SQL Compression Lab Project Results EXPERIENCE. INNOVATION. Velocity Proprietary and Confidential 17
  18. 18. Environment • • • • IBM 3850 M3 • 128gb RAM • 16 Cores SQL 2008 Enterprise x64 SP1 • 47.5gb memory allocated • Parallelism = 1 Storage • EMC San DMX 4 • Dual Attached • 64kb NTFS Allocation Size • 64kb Disk Offset Database • JDE_PRODUCTION (162gb) EXPERIENCE. INNOVATION. Velocity Proprietary and Confidential 18
  19. 19. Implementation • • • Restored JDE_PRODUCTION to SQL2008POC on SQL21 and created three databases • Uncompressed • Row Compression – 2 hours 9 minutes to compress • Page Compression – 2 hours 21 minutes to compress Captured a 1 hour long profile trace from SQL20 for JDE_PRODUCTION • 231 sql connections Replayed profile trace against all three of the databases and captured stats • SQL 2008 DMV’s • Windows Perfmon • Red-Gate EXPERIENCE. INNOVATION. Velocity Proprietary and Confidential 19
  20. 20. Database Size Database Sizes 180 160 162 140 Size in GB 120 ~90% Reduction in Storage Needed 100 80 64 60 40 16 20 0 Uncompressed Row Compression Page Compression Compression Type EXPERIENCE. INNOVATION. Velocity Proprietary and Confidential 20
  21. 21. Database Backup Size Backup Sizes 18 16 15.3 14 Size in GB 12 10 8 6.9 6 5 ~67% Reduction in Backup Storage Needed 4 2 0 Uncompressed Row Compression Page Compression Compression Type EXPERIENCE. INNOVATION. Velocity Proprietary and Confidential 21
  22. 22. Backup Times Backup Times 30 25 24 Minutes 20 ~93% Reduction in Backup Times 15 10 5.5 5 1.5 0 Uncompressed Row Compression Page Compression Compression Type EXPERIENCE. INNOVATION. Velocity Proprietary and Confidential 22
  23. 23. Restore Times Restore Times 30 25 24 Minutes 20 ~81% Reduction in Restore Times 15 10 5.5 5 1.5 0 Uncompressed Row Compression Page Compression Compression Type EXPERIENCE. INNOVATION. Velocity Proprietary and Confidential 23
  24. 24. Number of Reads Number of Reads 50000 45000 45671 40000 35000 Reads 30000 ~76% Reduction in Number of Reads 25000 20000 16238 15000 10721 10000 5000 0 Uncompressed Row Compression Page Compression Compression Type EXPERIENCE. INNOVATION. Velocity Proprietary and Confidential 24
  25. 25. Read Time Read Time (Seconds) 4500 4000 3995 3500 Seconds 3000 2500 ~93% Reduction in Read Time 2000 1500 1000 502 500 266 0 Uncompressed Row Compression Page Compression Compression Type EXPERIENCE. INNOVATION. Velocity Proprietary and Confidential 25
  26. 26. Number of Writes Number of Writes 21000 20792 20500 20272 20000 Writes 19500 19000 18386 18500 ~11% Reduction in Number of Writes 18000 17500 17000 Uncompressed Row Compression Page Compression Compression Type EXPERIENCE. INNOVATION. Velocity Proprietary and Confidential 26
  27. 27. Write Time Write Time (Seconds) 250 220 200 Seconds 165 154 150 ~24% Reduction in Write Time 100 50 0 Uncompressed Row Compression Page Compression Compression Type EXPERIENCE. INNOVATION. Velocity Proprietary and Confidential 27
  28. 28. CPU Utilization Percent CPU Time 17 16.80666781 16.38710853 Percent 16.5 16 15.5 ~8% Increase in Percent CPU Time Used 15.52862204 15 14.5 Uncompressed Row Compression Page Compression Compression Type EXPERIENCE. INNOVATION. Velocity Proprietary and Confidential 28
  29. 29. Memory Utilization Memory Utilization 2500000 2164773.926 MB Allocated 2000000 1500000 ~73% Reduction in Memory Needed 1000000 743887.2381 576674.3415 500000 0 Uncompressed Row Compression Page Compression Compression Type EXPERIENCE. INNOVATION. Velocity Proprietary and Confidential 29
  30. 30. Buffer Cache Hit Ratio Buffer Cache Hit Ratio 98 97.40602574 97.5 97 96.52317432 96.5 Percent 96 ~3% Increase in Buffer Cache Hit Ratio 95.5 95 94.5 94.46759604 94 93.5 93 92.5 Uncompressed Row Compression Page Compression Compression Type EXPERIENCE. INNOVATION. Velocity Proprietary and Confidential 30
  31. 31. Disk Utilization Avg. Disk sec/Transfer 0.025 0.021782055 Percent 0.02 0.014864164 0.015 0.015834584 ~33% Decrease in Disk Transfer Time 0.01 0.005 0 Uncompressed Row Compression Page Compression Compression Type EXPERIENCE. INNOVATION. Velocity Proprietary and Confidential 31
  32. 32. Top 5 LRQ - Duration Top 5 LRQ – Duration (Seconds) 200 180 180.92578 160 Seconds 140 120 100 86.04883 71.97558 80 ~60% Reduction in Duration 60 40 20 0 Uncompressed Row Compression Page Compression Compression Type EXPERIENCE. INNOVATION. Velocity Proprietary and Confidential 32
  33. 33. Top 5 LRQ – CPU Time (ms) Top 5 LRQ - CPU Time (Seconds) 60.00 50.00 49.43 40.44629 Seconds 40.00 33.87402 ~31% Reduction in CPU Time 30.00 20.00 10.00 0.00 Uncompressed Row Compression Page Compression Compression Type EXPERIENCE. INNOVATION. Velocity Proprietary and Confidential 33
  34. 34. Top 5 LRQ – Physical Reads Top 5 LRQ - Physical Reads 25000 20588 Physical Reads 20000 15000 ~77% Reduction in Physical Reads 10000 7063 4547 5000 0 Uncompressed Row Compression Page Compression Compression Type EXPERIENCE. INNOVATION. Velocity Proprietary and Confidential 34
  35. 35. Top 5 LRQ - Writes Top 5 LRQ - Writes 6000 5302 5000 4617 4204 Writes 4000 ~9% Reduction in Writes 3000 2000 1000 0 Uncompressed Row Compression Page Compression Compression Type EXPERIENCE. INNOVATION. Velocity Proprietary and Confidential 35
  36. 36. Notes • Recommended Indexes • Index’s are used more efficiently when the DB is compressed • Uncompressed – 40 new indexes were recommended • Row Compression – 26 new indexes were recommended • Page Compression – 16 new indexes were recommended EXPERIENCE. INNOVATION. Velocity Proprietary and Confidential 36
  37. 37. International Footprint to Serve Our Customers Glasgow, Scotland Seattle, Washington Minneapolis, Minnesota Tukwila, Washington London, Ontario New York, New York Denver, Colorado Temple Terrace, Florida EXPERIENCE. INNOVATION. Velocity Proprietary and Confidential 37
  38. 38. Velocity Is an Active Member of the Oracle® Community Built first hosted JD Edwards solution Support for Oracle edge apps and other partner software Expertise in JD Edwards, E-Business, Hyperion and others EXPERIENCE. INNOVATION. Customers with operations in 25 countries 40+ employees Oracle / JD Edwards certified Velocity Proprietary and Confidential 38
  39. 39. Questions? EXPERIENCE. INNOVATION. Velocity Proprietary and Confidential 39
  40. 40. Thank you. For more information: Paul.Shearer@Velocity.cc 803-386-9680 ©2013 Velocity Technology Solutions, Inc. EXPERIENCE. INNOVATION. EXPERIENCE. INNOVATION. Velocity Proprietary and Confidential 40

×