Get documents about "Keys to Optimizing Your Backup Environment Tivoli Storage Manager
Document Sample
Hosted by
Keys to Optimizing Your
Backup Environment:
Tivoli Storage Manager
John Merryman
GlassHouse Technologies, Inc.
Hosted by
Introduction
Audience Profile
Storage Management Interdependence
Backup Pain Points
TSM Architecture
• Server
• Storage
• Network
• Clients
Hosted by
Audience Response
Audience Profile
Your TSM Backup/Archive Environment
Number of Years using TSM
Number of TSM Administrators
Hosted by
Storage Management Interdependence
Storage and Production Backup
• Growing at 65% CAGR (Gartner)
• Consuming 48% of IT hardware budgets (Gartner)
Disaster Recovery
• Enabling Recovery – RPO/RTO impact
Archiving
• Compliance
“It’s no longer just
• Storage efficiency
backup, it’s data
• Disposal of data management.”
• Application Performance
Hosted by
TSM Pain Points
Business Problems Technology Problems
Backup reliability TSM architecture scalability
Overwhelming data growth Lack of TSM / OS tuning
Database and email data Network architecture
explosion Client-side issues (OS, antivirus,
Limited budgets and staff network)
Regulatory / data retention Increasingly complex requirements
requirements Shrinking backup windows
Lack of data management Disconnect between application
policies design and storage management
Poor operational practices
Solution Benefits
Optimize and right-size existing Lower TCO for TSM environment
infrastructure Proactive management model
Improve processes and Lowered risks and exposures
procedures
Scalable and manageable
Formalize TSM program environment
management
Hosted by
Business Problems
Backup reliability
Overwhelming data growth
Database and email data explosion
Limited budgets and staff
Regulatory / data retention requirements
Lack of data management policies
Poor operational practices
Hosted by
Technology Problems
• TSM architecture scalability
• Lack of TSM / OS tuning
• Network architecture
• Client-side issues (OS, antivirus, network)
• Increasingly complex requirements
• Shrinking backup windows
• Disconnect between application design and storage
management
Hosted by
Solution
• Optimize and right-size existing infrastructure
• Improve processes and procedures
• Formalize TSM program management
Hosted by
Benefits
• Lower TCO for TSM environment
• Proactive management model
• Lowered risks and exposures
• Scalable and manageable environment
Hosted by
TSM Pain Points
TSM Certifications, De-
TSM Client Administration Certifications, Re-Certifications
TSM Client Problem Resolution
8% 7%
2%
2% TSM Client Upgrades / Change
3% Management
2% TSM Client Process Improvement
2%
3%
Testing and Certifying New Client
Technologies
3%
Performing Restores
TSM Server / Device Problem
7%
Resolution
TSM Server Administration
TSM Server Monitoring, Reporting,
Planning
61% Testing and Certifying New Server
Technologies
Joint Effort Problem Resolution for
TSM Client Issues
Hosted by
TSM Architecture Overview
LAN
SAN
TSM
Server
Database
TSM Clients Recovery Log
Hosted by
TSM Architecture
Tape/SAN/Disk
SD
SD
HEWLETT
PACKARD
Technology TSM TSM
Server
1
10 00 BASE X
=
AMBER AC TIVI TY
GREEN LINK O K
=
DISABLED
FL ASHING GREEN =
2
5411 Enterprise Switch
SD
Clients
10/ 100 BASE TX
4 5 6 7 8 9 1 0 11 1 2 3 3 P OWER
1 2 3 3 MGMT
4 5 6 7 8 9 10 11 R 3R 3
Network
Hosted by
TSM Server Performance
Hardware
• CPU, Memory, HBAs
Server backplane/bus architecture
HBA location
Operating System Tuning
• VMTUNE for AIX has a significant impact
Must be tuned specifically for environment
• Asynchronous I/O (requires TSM settings too)
• Direct I/O (requires TSM settings too)
• Network options tuning
Hosted by
TSM Application Tuning
TSM Application Tuning
• Database Buffer Pool Sizing
Server Options bufpoolsize parameter
(up to 50%)
Selftunebufpoolsize parameter (up to 10%)
Relates closely to page space sizing
• Mirrorwrite db parallel and DBpageshadow
• Txngroupmax, movebatchsize, movesizethresh
Hosted by
TSM Application Tuning, II
64-bit and 32-bit code levels (and OS levels)
TSM performance tuning guide
• http://publib.boulder.ibm.com/tividd/td/TS
MM/SC32-9101-01/en_US/HTML/SC32-
9101-01.htm
Maintain a link between operating system,
application, and device performance analysis.
Hosted by
TSM Server Performance
Other Considerations
• Backup versioning policies have an effect.
• Schedule randomization versus static schedules
• Spread production workload across backup window.
• Control batch TSM processes to avoid overlap for
resource intensive processes.
• Low-level policies can have a high-level impact.
shrstatic versus shrdynamic
Changing retries settings
Hosted by
TSM Server Operations
Hosted by
Ideal TSM Batch Processing
/1 2 3
10 2/ 00
10:00
11:00
12:00
13:00
14:00
15:00
16:00
17:00
18:00
19:00
20:00
21:00
22:00
23:00
0:00
1:00
2:00
3:00
4:00
5:00
6:00
7:00
8:00
9:00
ac up TG oc s 60
B k S Pr es 3
ac up TG oc s 99
B k S Pr es 3
ac up TG oc s 00
B k S Pr es 4
at ba a up r es 72
D a se B ck P oc s 4
ig at n ro ss 5
M r io P ce 36
ig at n ro ss 3
M r io P ce 47
ig at n ro ss 4
M r io P ce 47
xp re nv nt y ro ss 9
E i I e or P ce 47
pa e cl ma on oc s 80
S c Re a ti Pr es 4
pa e cl ma on oc s 81
S c Re a ti Pr es 4
pa e cl ma on oc s 82
S c Re a ti Pr es 4
pa e cl ma on oc s 83
S c Re a ti Pr es 4
Hosted by
Poorly Optimized TSM Batch Processing
10/ 11/20 02 10/1 2/200 3 10/1 3/200 3 10/1 4/200 3
10
20
30
40
50
60
70
80
90
00
10
20
30
00
10
20
30
40
50
60
70
80
90
00
10
20
30
00
10
20
30
40
50
60
70
80
90
00
10
20
30
10
10
10
10
10
10
10
10
10
20
20
20
20
00
10
20
30
40
50
60
70
80
90
10
10
10
10
10
10
10
10
10
10
20
20
20
20
00
10
20
30
40
50
60
70
80
90
10
10
10
10
10
10
10
10
10
10
20
20
20
20
00
10
20
30
40
50
60
70
80
90
:
:
:
:
:
:
:
:
:
:
:
:
:
0
0
0
0
0
0
0
0
0
0
:
:
:
:
:
:
:
:
:
:
:
:
:
:
0
0
0
0
0
0
0
0
0
0
:
:
:
:
:
:
:
:
:
:
:
:
:
:
0
0
0
0
0
0
0
0
0
0
:
:
:
:
:
:
:
:
:
:
:
:
:
:
:
:
:
:
:
:
:
:
:
:
:
:
:
:
:
:
Backu p STG Proc ess 3 60
Migra tion Proce ss 36 5
Migra tion Proce ss 36 6
Migra tion Proce ss 36 7
Migra tion Proce ss 36 8
Migra tion Proce ss 36 9
Migra tion Proce ss 37 0
Migra tion Proce ss 37 1
Migra tion Proce ss 37 2
Migra tion Proce ss 37 3
Migra tion Proce ss 37 4
Migra tion Proce ss 37 5
Backu p STG Proc ess 3 76
Backu p STG Proc ess 3 77
Backu p STG Proc ess 3 78
Backu p STG Proc ess 3 79
Migra tion Proce ss 38 0
Migra tion Proce ss 38 1
Migra tion Proce ss 38 2
Migra tion Proce ss 38 3
Migra tion Proce ss 38 4
Migra tion Proce ss 38 5
Migra tion Proce ss 38 6
Migra tion Proce ss 38 7
Migra tion Proce ss 38 8
Migra tion Proce ss 38 9
Migra tion Proce ss 39 0
Migra tion Proce ss 39 1
Migra tion Proce ss 39 2
Migra tion Proce ss 39 3
Migra tion Proce ss 39 4
Migra tion Proce ss 39 5
Migra tion Proce ss 39 6
Migra tion Proce ss 39 7
Backu p STG Proc ess 3 98
Backu p STG Proc ess 3 99
Backu p STG Proc ess 4 00
Backu p STG Proc ess 4 01
Backu p STG Proc ess 4 02
Backu p STG Proc ess 4 03
Backu p STG Proc ess 4 04
Backu p STG Proc ess 4 05
Migra tion Proce ss 40 6
Migra tion Proce ss 40 7
Migra tion Proce ss 40 8
Migra tion Proce ss 40 9
Migra tion Proce ss 41 0
Migra tion Proce ss 41 1
Migra tion Proce ss 41 2
Migra tion Proce ss 41 3
Backu p STG Proc ess 4 14
Backu p STG Proc ess 4 15
Backu p STG Proc ess 4 16
Backu p STG Proc ess 4 17
Migra tion Proce ss 41 8
Migra tion Proce ss 41 9
Migra tion Proce ss 42 0
Migra tion Proce ss 42 1
Migra tion Proce ss 42 2
Migra tion Proce ss 42 3
Migra tion Proce ss 42 4
Migra tion Proce ss 42 5
Migra tion Proce ss 42 6
Migra tion Proce ss 42 7
Migra tion Proce ss 42 8
Migra tion Proce ss 42 9
Migra tion Proce ss 43 0
Migra tion Proce ss 43 1
Migra tion Proce ss 43 2
Migra tion Proce ss 43 3
Migra tion Proce ss 43 4
Migra tion Proce ss 43 5
Migra tion Proce ss 43 6
Migra tion Proce ss 43 7
Migra tion Proce ss 43 8
Migra tion Proce ss 43 9
Migra tion Proce ss 44 0
Migra tion Proce ss 44 1
Migra tion Proce ss 44 2
Migra tion Proce ss 44 3
Migra tion Proce ss 44 4
Migra tion Proce ss 44 5
Migra tion Proce ss 44 6
Datab ase B ackup Proc ess 4 47
Migra tion Proce ss 44 8
Migra tion Proce ss 44 9
Migra tion Proce ss 45 0
Migra tion Proce ss 45 1
Migra tion Proce ss 45 2
Migra tion Proce ss 45 3
Migra tion Proce ss 45 4
Migra tion Proce ss 45 5
Migra tion Proce ss 45 6
Migra tion Proce ss 45 7
Migra tion Proce ss 45 8
Migra tion Proce ss 45 9
Backu p STG Proc ess 4 60
Backu p STG Proc ess 4 61
Backu p STG Proc ess 4 62
Backu p STG Proc ess 4 63
Backu p STG Proc ess 4 64
Backu p STG Proc ess 4 65
Backu p STG Proc ess 4 66
Backu p STG Proc ess 4 67
Backu p STG Proc ess 4 68
Backu p STG Proc ess 4 69
Backu p STG Proc ess 4 70
Backu p STG Proc ess 4 71
Datab ase B ackup Proc ess 4 72
Migra tion Proce ss 47 3
Migra tion Proce ss 47 4
Migra tion Proce ss 47 5
Migra tion Proce ss 47 6
Migra tion Proce ss 47 7
Migra tion Proce ss 47 8
Expir e Inv entor y Pro cess 479
Space Recl amati on Pr ocess 480
Space Recl amati on Pr ocess 481
Space Recl amati on Pr ocess 482
Space Recl amati on Pr ocess 483
Space Recl amati on Pr ocess 484
Space Recl amati on Pr ocess 485
Space Recl amati on Pr ocess 486
Space Recl amati on Pr ocess 487
Datab ase B ackup Proc ess 4 88
Space Recl amati on Pr ocess 489
Space Recl amati on Pr ocess 490
Space Recl amati on Pr ocess 491
Space Recl amati on Pr ocess 492
Space Recl amati on Pr ocess 493
Space Recl amati on Pr ocess 494
Space Recl amati on Pr ocess 495
Space Recl amati on Pr ocess 496
Space Recl amati on Pr ocess 497
Space Recl amati on Pr ocess 498
Space Recl amati on Pr ocess 499
Datab ase B ackup Proc ess 5 00
Migra tion Proce ss 50 1
Migra tion Proce ss 50 2
Migra tion Proce ss 50 3
Migra tion Proce ss 50 4
Migra tion Proce ss 50 5
Migra tion Proce ss 50 6
Space Recl amati on Pr ocess 507
Space Recl amati on Pr ocess 508
Datab ase B ackup Proc ess 5 09
Space Recl amati on Pr ocess 510
Space Recl amati on Pr ocess 511
Datab ase B ackup Proc ess 5 12
Migra tion Proce ss 51 3
Migra tion Proce ss 51 4
Migra tion Proce ss 51 5
Migra tion Proce ss 51 6
Migra tion Proce ss 51 7
Migra tion Proce ss 51 8
Datab ase B ackup Proc ess 5 19
Space Recl amati on Pr ocess 520
Migra tion Proce ss 52 1
Migra tion Proce ss 52 2
Migra tion Proce ss 52 3
Migra tion Proce ss 52 4
Migra tion Proce ss 52 5
Migra tion Proce ss 52 6
Datab ase B ackup Proc ess 5 27
Backu p STG Proc ess 5 28
Backu p STG Proc ess 5 29
Backu p STG Proc ess 5 30
Backu p STG Proc ess 5 31
Backu p STG Proc ess 5 32
Backu p STG Proc ess 5 33
Backu p STG Proc ess 5 34
Backu p STG Proc ess 5 35
Migra tion Proce ss 53 6
Migra tion Proce ss 53 7
Migra tion Proce ss 53 8
Migra tion Proce ss 53 9
Migra tion Proce ss 54 0
Backu p STG Proc ess 5 41
Backu p STG Proc ess 5 42
Backu p STG Proc ess 5 43
Backu p STG Proc ess 5 44
Hosted by
TSM Architecture
Tape/SAN/Disk
SD
SD
HEWLETT
PACKARD
Technology TSM TSM
Server
1
1
10 00 BASE X
AMBER AC TIVI TY
GREEN =X
=
10 00 BASELINK O K
FL ASHING GREEN =
DISABLED
=
AMBER AC TIVI TY
=
GREEN LINK O K
2
2
54 Enterprise Switch
11
11
54 Enterprise Switch
SD
SD
Clients
DISABLED
FL ASHING GREEN =
10/ 100 BASE TX
4 5 6 7 8 9 1 0 11 1 2 3 3
POWER
10/ 100 BASE TX 1 2 3 3
MGMT
4 5 6 7 8 9 10 11 4 5 6 7 8 9 1 0 11 1 2 R
3 3 3R 3 POWER
1 2 3 3 MGMT
4 5 6 7 8 9 10 11 R 3R 3
Network
Hosted by
SAN Architecture
Disk Fabric Design
• Dedicated SAN/Disk resources for Database/Logs
• Dedicated SAN/Disk for storage pools
Tape Fabric Design
• Load Balancing and design is important
• Develop Ratios based on throughput and SAN fabric
1 GB Fabric w/ 3590-E1A Drives
• 1:3 Ratio HBA: Drive
2 GB Fabric w/ 3590-E1A Drives
• 1:6 Ratio HBA: Drive
Hosted by
Disk Architecture
Disk architecture
• Fast, redundant, flexible and scalable
File system design
• Create naming conventions and plan for scalability.
• Raw volumes versus file systems
• Create file system schema with disk layout in mind.
• Database/log performance (Random I/O)
• Storage volume performance (sequential)
Hosted by
Disk Architecture, II
Storage device architecture
• Design/layout is key for sequential versus random
disk performance requirements
• TSM is often the most I/O intensive application in
the datacenter.
• Disk performance tuning is often necessary.
Hosted by
Disk Storage Pools
Capacity Planning for disk storage pools
Plan for using Migration Delay for critical client
data.
• A note on storage pool caching and database
performance
Disk is traditionally used in TSM as a staging area
for daily migration to tape.
This paradigm is evolving.
Hosted by
Disk as a Primary Storage Pool
Considerations
• Detailed Capacity planning methods for disk storage
are required.
• TSM Backup Policies have to be carefully planned.
• Formatting and Recovering Terabytes of *.dsm files
is incredibly time and resource intensive.
• Performance-oriented disk storage pool design
becomes even more important.
• Copies to tape will still be required, and
simultaneous R/W can be a big issue for S-ATA
technologies.
Hosted by
Tape Architecture
Size and Type of Library
• Robot speed, Compression, Scalability, etc.
Number and Type of Tape Drives
• R/W Speed, Caching, Mount/Seek Times, etc.
Media Formats
• Create a media migration strategy and plan.
Consider emerging tape technologies.
• Tape is not dead and is not dying.
• High-end tape I/O is faster than low-end disk I/O.
• The real issue is random vs. sequential access to
data.
Hosted by
Tape Storage Pools
Stream large database objects data directly to tape.
• Lowers disk pool requirements
• Lowers TSM batch processing time
• Optimizes use of tape resources during backup window
• Schedule and resource planning is key.
• Resource Utilization and Maxnummp settings
Synchronous Copy Storage Pool Writes (“”)
• Can reduce batch window significantly
• Use only for large object backup clients.
Recovery optimized TSM tape storage pools
• Collocation=yes (+’s an –’s)
Hosted by
TSM Architecture
Tape/SAN/Disk
SD
SD
HEWLETT
PACKARD
Technology TSM TSM
Server
1
1
10 00 BASE X
AMBER AC TIVI TY
GREEN =X
=
10 00 BASELINK O K
FL ASHING GREEN =
DISABLED
=
AMBER AC TIVI TY
=
GREEN LINK O K
2
2
54 Enterprise Switch
11
11
54 Enterprise Switch
SD
SD
Clients
DISABLED
FL ASHING GREEN =
10/ 100 BASE TX
4 5 6 7 8 9 1 0 11 1 2 3 3
POWER
10/ 100 BASE TX 1 2 3 3 MGMT
4 5 6 7 8 9 10 11 4 5 6 7 8 9 1 0 11 1 2 R
3 3 3R 3 POWER
1 2 3 3 MGMT
4 5 6 7 8 9 10 11 R 3R 3
Network
Hosted by
Network Overview
Network Architecture
• Private / Public network configurations
Type of Architecture
• 100 Mb and Gigabit IP architectures are the defacto standard.
Develop Network Architecture Standards.
• Map network standards to SLAs and to service delivery.
• Publish metrics on network throughput.
• Enforce Standards.
Advancing a network architecture can be cost prohibitive
and time consuming…so can poor recovery times and data
loss.
Hosted by
Network Design
Dedicated Private VLAN architectures
• Logical network segregation and physical network
segregation
• Use jumbo frames and gigabit VLANs.
Improves streaming efficiencies to tape
9000 and 9004 MTU size differences
DNS Considerations
Multiple HBA’s for client network load balancing
• Etherchannel configurations
• Load balancing for network, switches, server HBAs
• Don’t use auto-negotiate network card settings.
Hosted by
TSM Architecture
Tape/SAN/Disk
SD
SD
HEWLETT
PACKARD
Technology TSM TSM
Server
1
1
10 00 BASE X
AMBER AC TIVI TY
GREEN =X
=
10 00 BASELINK O K
FL ASHING GREEN =
DISABLED
=
AMBER AC TIVI TY
=
GREEN LINK O K
2
2
54 Enterprise Switch
11
11
54 Enterprise Switch
SD
SD
Clients
DISABLED
FL ASHING GREEN =
10/ 100 BASE TX
4 5 6 7 8 9 1 0 11 1 2 3 3
POWER
10/ 100 BASE TX 1 2 3 3 MGMT
4 5 6 7 8 9 10 11 4 5 6 7 8 9 1 0 11 1 2 R
3 3 3R 3 POWER
1 2 3 3 MGMT
4 5 6 7 8 9 10 11 R 3R 3
Network
Hosted by
TSM Client Performance
Fact: Client machines are often sized for
application workload, not backup workload.
Client Tuning Parameters
• Resource Utilization (2-10)
• Memoryefficientbackup
• Virtualmountpoint
Network / Transaction Tuning Parameters
• Tcpnodelay
• Tcpwindowsize
• Tcpbufsize
• Txnbytelimit
Hosted by
TSM Client Configuration
File System Design and Type
• File system size and number impacts memory
consumption.
• TSM client can be memory intensive (300 MB
Memory / Million Objects).
Include/Exclude list size has a direct impact on
processing overhead.
• Leverage domain statements.
Hosted by
TSM Client Configuration, II
Journal-based backups
• Data drives only
• Journal size is limited to 2 GB
Image backups
LAN-free backups
Hosted by
TSM Client Performance
Backup Object Success Rates
Chart 2
1,000,000
100,000
Average Number of Objects
10,000
Average Objects Inspected
1,000 Average Objects Backed Up
Average Objects Failed
100
10
1
INEA2000
INEA2007
INEA2008
INEA2015
INEA2021
INAA2068
INAA2073
INDA0001
INDA0003
INDA2000
INDA2001
INDA2010
INDA2014
INPA2000
INSA2005
INSA2006
INPSWB01
Node Name
Hosted by
TSM Client Performance
Additional Considerations HE WLE TT
P ACK ARD
SD
• Data characteristics
Data type, volume, volatility
Number of files
• Client workloads and backup schedule
• Client Software Version
• Network load and utilization
• Communication with application architects
and developers is increasingly important.
Hosted by
TSM Client Performance
Client Versions
Ve rsion 3 Release 7 Level 1.0, 1 Clients
Version 3 Rele ase 1 Level 0.8, 1 Clients
Version 5 Release 1 Level 6.0, 1 Clie nts
Ver sion 4 Re lease 1 Level 0.0, 2 Clients
Version 3 Release 1 Level 0.3, 3 Clients
Version 3 Release 1 L evel 0.1, 3 Cli ents
Version 3 Release 1 Level 0.7, 4 Clients
Version 4 Release 2 Level 0.0, 4 Clie nts
Version 4 Release 2 Level 1.0, 8 Clients Ve rsion 4 Re lease 1 Level 2.0, 45 Clients
Version 4 Release 1 Level 1 .0, 11 Clients
Version 5 Release 1 Level 5.14, 12 Clients
(?), 14 Clien ts
V ersion 5 Release 1 Level 5.0, 27 Clients
V ersion 3 Release 1 Level 0.5, 21 Clients
Version 4 Release 1 Level 2.12, 24 Cli ents
Hosted by
To receive a copy of the charts seen in this
presentation and more for your
environment go to:
www.glasshouse.com/backup
For a “cheat sheet” see the
highlighter in your conference bag.
Hosted by
Thank you.
Questions?
Mr. Merryman will be available at the
Ask-the-Experts booth in the Exhibit Hall:
Tuesday 4-5 PM
Hosted by
Information. Optimized.
GlassHouse Technologies, Inc.
200 Crossing Boulevard
Framingham, MA 01702
www.glasshousetech.com jmerryman@glasshouse.com
Related docs
