Transcript Document

Virtualizing Microsoft Exchange Server 2010 & 2013
About Me – Charles Derber
Work as consultant on Microsoft Technologies (Active Directory & Exchange Server )
Have overall experience of 7 years in the IT industry
Have been working on different versions:
 Microsoft Exchange Server 2000 / 2003 / 2007 / 2010 / 2013(not yet on production)
I’m Passionate about Microsoft AD & Exchange(specific to), I do blog @
http://charlesgate86.wordpress.com
Agenda
 Virtualizing Exchange why ?
 Exchange Supportability
 Exchange Best Practices for Virtualization
 Exchange High Availability Options
 Planning for Exchange virtualization
 Disaster recovery & Backup consideration
 Deployment Recommendations
 Exchange Server guidelines for Virtualization
 Using Mailbox Calculator for planning Exchange role
Should I virtualize…?
 If you don’t have a good reason to virtualize Exchange don’t do it – simplicity rules.
 Exchange is not build to be a virtualized product platform but to work great on Windows and
provide fantastic email collaboration solution
 Exchange when virtualized the objective is to get something out of it – like taking advantage of
live migration, maintenance of physical host, etc.
Virtualizing Exchange why….?
Cost
 OS licenses ( No. of virtual machine OS / Standard, Enterprise or Datacenter Edition)
 Varies and very specific for business requirement
Server Consolidation –
 Power, Space & Cooling
Maximum utilization of available physical resources
 Processor & Memory
Virtualizing Exchange why….?
 Microsoft Hyper V – the most tested Hypervisor for Exchange
Thousands of Exchange test machines are run on Hyper V everyday
Feedback loop between Exchange & Hyper V – Its Microsoft.
Exchange Virtualization Supportability
Host Operating Systems supported
 Windows Server 2008 R2 & 2012)
 Hyper V (2008 R2 & 2012)
Third party hypervisor (SVVP) – Vmware / XEN etc.
Guest Operating Systems supported
 Exchange 2010 – Windows 2008 SP2 or R2, 2012(with Exchange 2010 Sp3)
 2010 RTM – All roles except Unified Messaging
 2010 with Sp1 – UM supported
 Exchange 2013 – Windows 2008 R2 with Sp1, 2012
 RTM – All roles (CAS + MBX)
Exchange Virtualization Supportability – SVVP
Unified Messaging Virtualization
 Requires Exchange 2010 SP1+ for support
Enabled high quality real-time audio processing in a virtual deployment, and in the SP1 release of Exchange
2010 team integrated those changes into the UM role
 VM should have a minimum of:
– (4) Virtual Processors/Cores
– Requires 1:1 vCore:pCore allocation
– 16GB of memory
 Unified Messaging server role must be stand alone (2010) – Hub, CAS, and Mailbox can not be
in the same VM as Unified Messaging role
Virtual Machine Guest - Supported
Maximum CPU ratio of 2:1 vCPU:pCPU –
 Example: If you have 8 cores, maximum supported is 16 vCPUs assigned to guests –
Recommended ratio is 1:1, don’t overcommit CPU in production
iSCSI initiator in guest is supported –
 Check with your vendor, make sure VM network stack supports jumbo frames, full network
fidelity
Virtual Machine Guest - Supported
Exchange 2013 supports VHD storage on SMB 3.0 file shares
Can be shares presented from Windows Server 2012 or other implementations of SMB 3.0
Specific to VHD storage – no direct access to shares from Exchange
Virtual Machine Guest – Not Supported
Dynamic memory & memory over-commit –
 Not supported for any Exchange role
Hypervisor snapshots (Time Travel) –
 Not supported for any Exchange role
Differencing/delta/dynamic disks –
 Not supported for any Exchange role
Other applications installed on the root –
 Only deploy your management, monitoring, A/V, etc.
Virtual Machine Guest – Not Supported
 Hyper V Replica provides DR for “any” VM via log shipping to a remote hypervisor - not
supported for Exchange
 Replica makes sense for apps that don’t have DR capability built-in to the product
 Use DAG to obtain better HA & DR with Exchange 2010 & 2013
Host-Based High Availability (Live
Migration & vMotion, etc)
 Exchange 2010 RTM : Can not mix with DAG
 Exchange 2010 SP1 + : Can mix with DAG*
 Exchange 2013 : Can mix with DAG*
Host-Based High Availability
What is Host-Based High Availability?
 Automatic failover of virtual machines to another virtualization host in the event of a critical hardware
failure (virtualization platform independent)
Key Concepts –
 Cold boot: Bringing a system from a “power-off” state to a clean, fresh start of the OS
 Saved state: Many hypervisors allow you to save state, or “hibernate” a VM rather than shutting it
down or turning it off.
Host-Based High Availability – cont’d
What you need to be aware of:
 Not an Exchange-aware solution – No knowledge of transaction logs, clean/dirty database
dismount, database checksums, Exchange-health, etc.
 Only protects against server hardware/network failure
 No protection against storage failure / data corruption
 Requires the guest VM to perform a cold boot (this is HA?)
Host-Based High Availability – cont’d
What is Supported:
Automatically failing VMs to an alternate cluster node in the event of a critical host hardware
issue
 Important! VM must come up from a cold start, not saved state or warm start
Live Migration / vMotion / etc.
What is Not Supported:
Quick Migration (Windows Server 2008 pre-R2)
Anything that pauses/saves state, migrates, and then resumes
“Time Traveling”
Which HA to choose?
Microsoft recommends Exchange HA (DAG)
 Exchange-aware HA (log shipping, single page restore, no more -1018s, best copy selection,
etc.)
 Understand “failure domains”
 Power, network, storage, rack, blade chassis, etc…
Sizing Recommendations – CPU
 Plan to add additional CPU overhead on Exchange VMs
◦ ~10% for Hyper-V
◦ Follow vendor guidance for 3rd-party
 Best way to size for this is using Role Requirements Calculator
◦ Note “Server Role Virtualization” and “Hypervisor CPU Adjustment Factor”
Sizing Recommendations – Other Resources
 Memory generally not impacted
◦ Exchange is not NUMA aware, but will take advantage of OS optimizations
 Storage should be optimized for IO latency and high availability
 Take advantage of hypervisor networking flexibility to provide availability and performance
 In general: size using guidance for physical, apply to virtual
Sizing Recommendations – with other VMs
 Use reservation options to ensure that Exchange gets the resources it needs
◦ Allows other workloads to take advantage of overcommit
 As a best practice, never overcommit any resources in a way that could impact Exchange VMs
Exchange VM placement recommendation
Use common sense when placing your VMs
 Deploy VMs with the same role across multiple root
 Do NOT deploy MBX VMs in the same DAG on the same root server
Exchange VM placement recommendation
Placing any dependencies of Exchange on the same infrastructure also impacts availability
◦ Active Directory / Global Catalog
◦ Witness servers
 Placing multiple mailbox DB copies on the same infrastructure impacts availability
The Problem – Memory
 Hyper-V’s Dynamic Memory and VMware’s Ballooning are fantastic for lab
environments
 For production Exchange servers, just don’t do it!
 Exchange code doesn’t deal well with disappearing memory
 End result for the mailbox role is cache served from the page file or very small
DB cache
 Best practice: static memory configuration for Exchange VMs
The Problem – Dynamic Memory
The Problem – Migration
Adjust cluster heartbeat if necessary
 Target less than 5 seconds for Live Migration / vMotion
 Adjust if Live Migration takes longer than 5 seconds, but no more than 10 seconds
Enable jumbo frames on the Live Migration network(s)
On the live migration network, change receive buffers to 8192 (default for the network
interfaces in test was 896) on each host.
Use very fast networks (4Gb, 10Gb)
Best Practices –– Virtual Networking
 Allocate separate network adapters/networks for Vmotion/Live migration,
VMware FT logging traffic, and ESX/Hyper V console access management.
 Allocate at least two network adapters for Exchange production traffic to
leverage Vmware/Hyper V NIC teaming capabilities. Generally, at least four
network adapters are recommended per ESX host/Hyper V.
Ensure that the latest patches for the hypervisor are deployed to ensure
optimal performance.
KB 2517329. Performance decreases in Windows Server 2008 R2 when the
Hyper-V role is installed on a computer that uses Intel Westmere or Sandy
Bridge processors
KB 2000977. Hyper-V: Performance decrease in VMs on Intel Xeon 5500
(Nehalem) systems
Disaster Recovery / Backup
Considerations
Design your backup solution carefully!
 Use Exchange-aware VSS solutions in guests to get the Exchange stores and guest OS instance
 Use VSS on the host to backup the hypervisor and configuration
Interesting DR scenarios
 Some physical, some virtual
 Ensure that if you have a virtual machine as a target for multiple physical server in your DR scenario,
that it can handle the worst case DR
Hardware Considerations
Disk subsystem must still support required IOPS – Jetstress is still the gold standard
Test as a system
 Jetstress and LoadGen all VMs on the system to stress the entire system with a realistic peak load
 Unless your clients are going to be virtual in production (VMware VDI for instance), test with physical
LoadGen clients
Watch “over-oversubscribe” of processors – Remember the vCore : pCore ratio (2:1 max
supported, 1:1 recommended)
Hardware Considerations – Cont’d
Network
 Bandwidth
 Take into account iSCSI when necessary!
 High bandwidth customers should avoid 1Gb iSCSI
 Newer FC4Gb, FC8Gb, 10Gb iSCSI or FCoE better
New hardware vs reuse/upgrade of old
 Will older procs support 64-bit guests (watch new procs, too)
 Can you upgrade the procs in an old server to work?
 Will upgrading the memory cost more than a new server?
Deployment Recommendations
Exchange Server 2010/2013 is not “virtualization aware”
Core Exchange Design Principles Still Apply
 Design for Performance, Reliability and Capacity
 Design for Usage Profiles (CAS/MBX)
 Design for Message Profiles (Hub/Edge)
Virtualization Design Principles Now Apply
 Design for Performance, Reliability and Capacity
 Virtual machines should be sized specific to the Exchange role (EDGE, HUB, CAS, MBX, multi-role)
 Hosts should be sized to accommodate the guests that they will support
Mailbox Server Guidelines
Database Cache requirements are the same for physical and virtual deployments
Mailbox Server Guidelines
Hypervisor and the Virtualization Stack consume CPU
Mailbox Server Guidelines (2013)
Expect that the 10% rule of thumb will continue
Mailbox Storage Configuration
Virtual SCSI (passthrough or fixed disk VHD/VMDK)
 Recommended configuration for database and log volumes
iSCSI
 Standard best practice for iSCSI connected storage apply (dedicated NIC, jumbo frames, offload, etc.)
 iSCSI initiator in the guest is supported but need to account for reduced performance
 Clarify support between hypervisor vendor and storage vendor (certain combinations are not supported)
Mailbox Server Role Requirements Calculator
 Created during Exchange 2007 product cycle
 The ‘storage calculator’ – but much, much more now!
 Properly calculates host CPU, memory, storage requirements, network bandwidth for
replication
 Newer versions generate database/dag deployment scripts
Exchange Server Role Requirements Calculator – Version 20.8
Exchange Server Role Requirements Calculator
 Latest Mailbox Server Role Requirements Calculator (v17.2) now includes support for
virtualization!
 Hypervisor CPU Adjustment Factor is the virtualization overhead (rule of thumb is 10%)
Exchange Server Role Requirements Calculator
 When virtualizing Exchange on Hyper-V, you can expect a 10% CPU overhead on the
guest machine for Hyper-V. Other hypervisors may provide different guidance for CPU
overhead
 For example, let’s say I am deploying a processor model Intel Xeon X5470 system which
includes four sockets, each containing an 8-core processor, then the SPECInt2006 rate
value for the system is 147.
Exchange Server Role Requirements Calculator
Exchange Processor Query Tool
Get the Exchange Processor Query Tool –
http://msexchangeteam.com/archive/2010/10/27/456738.aspx
This allows you to put in a processor and number of cores, and have the tool give you the average
SpecInt value you can use in the calculator
Exchange Processor Query Tool
Now that we’ve talked about virtualizing Exchange…
WHY NOT VIRTUALIZE?
Why NOT Virtualize Exchange?
Virtualization does add complexity – additional deployment
Virtualization does impact performance – you are adding additional layer between hardware & Exchange
You do still have to manage those virtual servers as if they were physical
 Licensing, Hotfixes, Service Packs, Updates, Monitoring, etc.
You also have to “manage” that virtualization platform
Are there alternatives? Do they make sense?
 Can a physical architecture make full use of hardware?
 Would use of blade server technologies make more sense than virtualization?
The Alternative: Exchange Multi-Role
Server
 Current Microsoft guidance is to deploy multi-role servers (CAS + HUB + MBX – 2010 & CAS +
MBX – 2013) instead of virtualization
 Take full advantage of the physical machine
Additional reading…
Windows Server 2008 R2 Hyper-V & 2012
http://www.microsoft.com/windowsserver2008/en/us/hyperv-main.aspx
http://technet.microsoft.com/en-us/library/cc732470(v=ws.10).aspx
Windows Virtualization Team Blog
http://blogs.technet.com/virtualization
Infrastructure Planning and Design Guides for Virtualization
http://technet.microsoft.com/en-us/solutionaccelerators/ee395429.aspx?SA_CE=VIRT-IPD-WEB-MSCOM-2009-09-21
Exchange Server Guidance
http://technet.microsoft.com/en-us/library/bb124558(EXCHG.140).aspx
Exchange Team Blog
http://blogs.technet.com/exchange
Best Practices Whitepaper for Virtualizing Exchange 2010 & 2013
http://www.microsoft.com/downloads/en/details.aspx?FamilyID=8647c69d-6c2c-40ca-977e-18c2379b07ad
http://download.microsoft.com/download/4/A/C/4AC32FD3-220E-45DC-AA97-DBDBE19C15B2