Exchange Server 2013 Sizing

Report
http://ignite.office.com
Explain the Exchange 2013 sizing process
Describe what has changed since prior releases
Demonstrate use of the calculator
Discuss other tools & resources for sizing
Exchange Dogfood, MSIT, Partner, Customer & field feedback
Capacity planning
Sizing guidance
Performance bug detection
Check out Exchange Server\V15\Logging\Diagnostics\DailyPerformanceLogs
Consider extra safety margins when targeting “max” CPU
Minimize overdeployment
Read/understand
sizing, scalability,
capacity guidance
• Technet Documentation
• Exchange team blog
• Partner sizing guidance
Collect data on
existing deployment
• User profile
(messages
sent+received per
day)
• Average message
size
Define constraints
based on customer
requirements
•
•
•
•
•
•
•
# of copies
Backup requirements
Storage architecture
SafetyNet duration
Virtualization
Growth plans
3rd party products
Input profile data and
design constraints
into calculator tool
(or calculate
manually)
• Always use the latest
calculator
Consider impact of
various options
provided by sizing
results
• Cost
• Rebuild times
• Impact on high
availability
Finalize design
• Storage calculator
provides
configuration scripts
• Archive the
calculator as
documentation of
the sizing process
Enterprise network
Similar to Exchange 2010 multi-role
External
SMTP
servers
Edge transport
Routing and AV/AS
CAS array
DAG
CAS
MBX
CAS
MBX
CAS
MBX
CAS
MBX
CAS
MBX
Layer 4LB
Consider hardware platforms that
provide the right balance of resources
Forefront
online
Protection for
Exchange
Mobile
phone
Web
browser
Outlook
(remote user)
Outlook (local user)
Line of
business
application
AD
Phone system
(PBX or VOIP)
8GB minimum RAM
4GB minimum RAM
Minimum Disk Space (GB)
30GB
100
10
1
Exchange 2003
0.1
Exchange 2007
Exchange 2010
Exchange 2013
Simplified deployment & connectivity model
Cache efficiencies
Hardware efficiencies (balanced resource utilization)
Unit of scale for capacity planning
Tradeoffs result in some increased resource usage
Cache sizing is different
Everything interacts (and workload management mediates)
Managed availability has a measurable impact on the system
New content indexing architecture impacts performance
Unified Messaging enabled on all Mailbox servers
Method for computing space requirement similar to Exchange 2010, with some important changes
Plus space for additional index set per volume (master merge)
Note that impact of space for master merge is reduced with multiple DBs per-volume
We have seen higher reduction in various tests, guidance is conservative and based on production
observations
Checkpoint depth is now consistent for all scenarios, so IOPS requirements are the same
Database IOPS
10
1
0.1
0.01
50
100
150
200
250
300
350
400
Messages Sent/Received Per-User Per-Day
Exchange 2003
Exchange 2007
Exchange 2010
Exchange 2013
450
500
Guidance shows method for calculating capacity requirements
Low IOPS suggest that placing transport queue on system/install volume is now feasible in many
scenarios
Significant transport throughput benefit seen from a protected write cache disk controller, set to
100% write cache
Per-passive multiplier on the active has been
removed in 2013
50
2.13
2.66
100
150
200
250
300
350
400
450
500
4.25
6.38
8.50
10.63
12.75
14.88
17.00
19.13
21.25
5.31
7.97
10.63
13.28
15.94
18.59
21.25
23.91
26.56
0.69
1.37
2.06
2.74
3.43
4.11
4.80
5.48
6.17
6.85
Processor throughput can potentially reach additional 20-30%, but not necessarily
Significant impact to some Exchange service memory footprints – in some cases doubling the
memory footprint
Uses common heap and cleanup process (can be concurrent or not)
Allocates a heap and thread per logical proc
Hyperthreading “doubles” the logical processors in the system, and thus increases memory
requirements
50
12
100
24
150
36
200
48
250
60
300
72
350
84
400
96
450
108
500
120
2GB +
2GB ×
worst−case active DBs per−server × users per−DB × mbx mcycles per−user × 0.25
per−core mcycles
1-10
11-20
21-30
31-40
41-50
8
10
12
14
16
Cost has dropped, many customers are standardizing on 10GB Ethernet in their datacenters
Note that UM in 2013 is a component of the Mailbox role, may need to adjust user distribution to
optimize UM utilization/concurrency
Stateless
Connection scalability
Low CPU & memory footprint
Load balancing optimizations
Namespace optimizations
Low resource utilization makes multi-role deployment (or virtualization) attractive
CAS is a net-new role in 2013, adding performance “cost”
Shift of processing resources from LB layer to CAS may negate new resource demand
CAS sizing
Transport storage sizing
Multiple databases per-volume (JBOD) support
High availability architecture improvements
http://aka.ms/E2013Calc
Demo
Exchange 2013 Role Requirements Calculator
http://aka.ms/E2013Calc
http://aka.ms/Exchange2013SizingGuidanceBlog
http://aka.ms/ExProcQueryTool
Event log captured
Errors associated with specific volumes
Threads controlled globally instead of per-DB, better automatic tuning
http://aka.ms/Jetstress2013
Support for protocol & connection changes in Exchange 2013
Stability
Many bug fixes
Explain the Exchange 2013 sizing process
Describe what has changed since prior releases
Demonstrate use of the calculator
Discuss other tools & resources for sizing
Can my system handle the desired workload?
What is the impact of a specific features or configuration?
How many users of a given profile can this hardware support?

similar documents