1. Why Hyper-V?
Competitive Advantages of Windows Server 2012 Hyper-V & Microsoft Hyper-V
Server 2012 over the VMware vSphere Hypervisor and vSphere 5.1
Matt McSpirit
Technical Product Marketing Manager
Microsoft Corporation
Microsoft Virtual Academy
3. Before Windows Server 2012 Hyper-
V
June 2008
H
yper-V
introduced
in
W
indow
sServer2008
October 2008
H
yper-V
Server2008
launched
October 2009W
indow
sServer2008
R2
H
yper-
V
&
H
yper-V
Server2008
R2
launched
Live Migration
Cluster Shared Volumes
Processor Compatibility
Hot-Add Storage
Performance &
Scalability
Improvements
February 2011
SP1
forW
indow
sServer2008
R2
&
Hyper-V
Server2008
R2
launched
Dynamic Memory
RemoteFX
6. Hyper-V Scalability
Improvements
“Nothing
from
Microsoft,
and I mean
literally
nothing has
ever been
this
ambitious”
— Jason Perlow, ZDNet
System Resource
Hyper-V
(2008 R2)
Hyper-V
(2012)
Improvement
Factor
Host
Logical Processors 64 320 5×
Physical Memory 1TB 4TB 4×
Virtual CPUs per Host 512 2,048 4×
VM
Virtual CPUs per VM 4 64 16×
Memory per VM 64GB 1TB 16×
Active VMs per Host 384 1,024 2.7×
Guest NUMA No Yes -
Cluster
Maximum Nodes 16 64 4×
Maximum VMs 1,000 8,000 8×
7. VMware Comparison
7
Only Hyper-V
allows the
creation of
VMs with up
to 64 vCPUs
& 1TB RAM,
in all
editions
System Resource
Hyper-V
(2012)
vSphere Hypervisor
vSphere 5.1
Enterprise Plus
Host
Logical Processors 320 160 160
Physical Memory 4TB 32GB1 2TB
Virtual CPUs per Host 2,048 2,048 2,048
VM
Virtual CPUs per VM 64 8 642
Memory per VM 1TB 32GB1 1TB
Active VMs per Host 1,024 512 512
Guest NUMA Yes Yes Yes
Cluster
Maximum Nodes 64 N/A3 32
Maximum VMs 8,000 N/A3 4,000
1 Host physical memory is capped at 32GB thus maximum VM memory is also restricted to 32GB usage.
2 vSphere 5.1 Enterprise Plus is the only vSphere edition that supports 64 vCPUs.
Enterprise edition supports 32 vCPU per VM with all other editions supporting 8 vCPUs per VM
3 For clustering/high availability, customers must purchase vSphere
vSphere Hypervisor / vSphere 5.x Ent+ Information: http://www.vmware.com/pdf/vsphere5/r51/vsphere-51-configuration-maximums.pdf, https://
www.vmware.com/files/pdf/techpaper/Whats-New-VMware-vSphere-51-Platform-Technical-Whitepaper.pdf and http://www.vmware.com/products/vsphere-hypervisor/faq.html
8. Enhanced Storage Capabilities
Virtual Fibre
Channel
Connect a
VM directly
to FC SAN
without
sacrificing
Native 4K
Disk Support
Take
advantage of
enhanced
density and
reliability
64TB Virtual
Hard Disks
Increased
capacity,
protection &
alignment
optimization
9. Enhanced Storage Capabilities
Boot from
USB Disk
Flexible
deployment
option for
diskless servers
(Hyper-V
Offloaded
Data Transfer
Offloads
storage-
intensive
tasks to the
SAN
Storage
Spaces
Storage
resiliency and
availability
with
commodity
11. VMware Comparison
11
Hyper-V
allows the
creation of
virtual disks
that are 32
times the
size of those
created on
vSphere 5.1
Capability
Hyper-V
(2012)
vSphere Hypervisor
vSphere 5.1
Enterprise Plus
Virtual Fiber Channel Yes Yes Yes
3rd Party Multipathing (MPIO) Yes No Yes (VAMP)1
Native 4-KB Disk Support Yes No No
Maximum Virtual Disk Size 64TB VHDX 2TB VMDK 2TB VMDK
Maximum Pass Through Disk Size 256TB+2 64TB 64TB
Offloaded Data Transfer Yes No Yes (VAAI)3
Boot from USB Disk Yes4 Yes Yes
Storage Pooling Yes No No
1 vStorage API for Multipathing (VAMP) is only available in Enterprise & Enterprise Plus editions of vSphere 5.1
2 The maximum size of a physical disk attached to a virtual machine is determined by the guest operating system
and
the chosen file system within the guest. More recent Windows Server operating systems support disks in excess of
256TB in size
3 vStorage API for Array Integration (VAAI) is only available in Enterprise & Enterprise Plus editions of vSphere 5.1
4 Hyper-V Server 2012 Only
vSphere Hypervisor / vSphere 5.x Ent+ Information: http://www.vmware.com/pdf/vsphere5/r51/vsphere-51-configuration-maximums.pdf and http://
www.vmware.com/products/vsphere/buy/editions_comparison.html
12. Enhanced Resource Management
Dynamic
Memory
Increased
control for
greater
virtual
machine
Resource
Metering
Track
historical
data for
virtual
machine
Quality of
Service (QoS)
Consistent
level of
performance
based on
SLAs
Data Center
Bridging
Converge
network
traffic to
provide
enhanced
14. VMware Comparison
Hyper-V
provides QoS
in all editions,
unlike
VMware,
where QoS is
only
available in
the
Enterprise
Plus edition
Capability
Hyper-V
(2012)
vSphere Hypervisor
vSphere 5.1
Enterprise Plus
Dynamic Memory Yes Yes Yes
Resource Metering Yes Yes1 Yes
Quality of Service Yes No Yes2
Data Center Bridging (DCB) Yes Yes Yes
1 Without vCenter, Resource Metering in the vSphere Hypervisor is only available on an individual host by host basis.
2 Quality of Service (QoS) is only available in the Enterprise Plus edition of vSphere 5.1
vSphere Hypervisor / vSphere 5.x Ent+ Information: http://www.vmware.com/pdf/vsphere5/r51/vsphere-51-configuration-maximums.pdf and
http://www.vmware.com/products/vsphere/buy/editions_comparison.html
16. Hyper-V Extensible Switch
PVLANS
ARP/ND
Poisoning
Protection
DHCP Guard
Protection
Virtual Port
ACLs
Trunk Mode
to Virtual
Machines
Monitoring &
Port
Mirroring
Windows PowerShell & WMI Management
16
The Hyper-V
Extensible
Switch allows a
deeper
integration
with customers’
existing
network
infrastructure,
monitoring and
security tools
17. Hyper-V Extensible Switch
Cisco
Nexus 1000V
UCS VM-FEX
5nine
Security
Manager
NEC
OpenFlow
InMon
sFlow
Multiple Partner Extensions
17
Hyper-V
Extensible
Switch is an
open platform
that lets
multiple
vendors
provide
extensions that
are written to
standard
Windows API
Packet
Inspection
Packet
Filtering
Network
Forwarding
Intrusion
Detection
18. VMware Comparison
The Hyper-V
Extensible
Switch is
open and
extensible,
unlike
VMware’s
vSwitch, which
is closed,
and
Capability
Hyper-V
(2012)
vSphere
Hypervisor
vSphere 5.1
Enterprise Plus
Extensible vSwitch Yes No Replaceable1
Confirmed Partner Extensions 5 No 2
Private Virtual LAN (PVLAN) Yes No Yes1
ARP Spoofing Protection Yes No vCNS/Partner2
DHCP Snooping Protection Yes No vCNS/Partner2
Virtual Port ACLs Yes No vCNS/Partner2
Trunk Mode to Virtual Machines Yes No Yes3
Port Monitoring Yes Per Port Group Yes3
Port Mirroring Yes Per Port Group Yes3
1 The vSphere Distributed Switch (required for PVLAN capability) is available only in the Enterprise Plus edition of
vSphere 5.1 and is replaceable (By Partners such as Cisco/IBM) rather than extensible.
2 ARP Spoofing, DHCP Snooping Protection & Virtual Port ACLs require the App component of VMware vCloud
Network & Security (vCNS) product or a Partner solution, all of which are additional purchases
3 Trunking VLANs to individual vNICs, Port Monitoring and Mirroring at a granular level requires vSphere Distributed
Switch, which is available in the Enterprise Plus edition of vSphere 5.1
vSphere Hypervisor / vSphere 5.x Ent+ Information: http://www.vmware.com/products/cisco-nexus-1000V/overview.html,
http://www-03.ibm.com/systems/networking/switches/virtual/dvs5000v/, http://
www.vmware.com/technical-resources/virtualization-topics/virtual-networking/distributed-virtual-switches.html, http://
www.vmware.com/files/pdf/techpaper/Whats-New-VMware-vSphere-51-Network-Technical-Whitepaper.pdf, http://www.vmware.com/products/vshield-app/features.html and
http://www.cisco.com/en/US/prod/collateral/switches/ps9441/ps9902/data_sheet_c78-492971.html
19. Networking Performance
Dynamic
VMq
IPsec Task
Offload
SR-IOV
Support
The Hyper-V
Extensible
Switch takes
advantage of
hardware
innovation to
drive the
highest
levels of
networking
performance
within virtual
Dynamically span multiple CPUs when processing
virtual machine network traffic
Offload IPsec processing from within virtual machine,
to physical network adaptor, enhancing performance
Map virtual function of an SR-IOV capable physical
network adaptor, directly to a virtual machine
20. Physical Security
BitLocker
ensures your
data stays
secure, even
when your
Hyper-V hosts,
clusters, and
storage reside
in less-
physically-
secure
locations
Local Disk
Traditional
Cluster Disk CSV 2.0
BitLocker
21. VMware Comparison
Unlike VMware,
Hyper-V’s SR-
IOV support
ensures the
highest
performance
without
sacrificing
key features
such as Live
Migration
Capability
Hyper-V
(2012)
vSphere
Hypervisor
vSphere 5.1
Enterprise Plus
Dynamic Virtual Machine Queue Yes NetQueue1 NetQueue1
IPsec Task Offload Yes No No
SR-IOV with Live Migration Yes No2 No2
Storage Encryption Yes No No
1 VMware vSphere and the vSphere Hypervisor support VMq only (NetQueue)
2 VMware’s SR-IOV implementation does not support vMotion, HA or Fault Tolerance. DirectPath I/O, whilst
not
identical to SR-IOV, aims to provide virtual machines with more direct access to hardware devices, with
network cards being a good example. Whilst on the surface, this will boost VM networking performance, and
reduce the burden on host CPU cycles, in reality, there are a number of caveats in using DirectPath I/O:
•
Very small Hardware Compatibility List
•
No Memory Overcommit
•
No vMotion (unless running certain configurations of Cisco UCS)
•
No Fault Tolerance
•
No Network I/O Control
•
No VM Snapshots (unless running certain configurations of Cisco UCS)
•
No Suspend/Resume (unless running certain configurations of Cisco UCS)
•
No VMsafe/Endpoint Security support
SR-IOV also requires the vSphere Distributed Switch, meaning customers have to upgrade to the highest
vSphere edition to take advantage of this capability. No such restrictions are imposed when using SR-IOV in
Hyper-V, ensuring customers can combine the highest levels of performance with the flexibility they need for
an agile infrastructure.
vSphere Hypervisor / vSphere 5.x Ent+ Information: http://www.vmware.com/pdf/Perf_Best_Practices_vSphere5.1.pdf
23. Virtual Machine Mobility
Live
Migration
Faster,
unrestricted,
simultaneous
VM live
migrations
between cluster
Live Storage
Migration
Move the virtual
hard disks of
running virtual
machines to a
different storage
location with no
Shared-Nothing
Live Migration
Move Virtual
Machines
between Hyper-
V hosts with
nothing but a
network cable
25. Network Virtualization
Secure
Isolation
Isolate network
traffic from
different
business units or
customers on a
shared
Flexible
Migrations
Move VMs as
needed within
your virtual
infrastructure
while preserving
their virtual
Seamless
Integration
Transparently
integrate these
private
networks into a
preexisting
infrastructure
26. VMware Comparison
Only Hyper-V
provides key
VM migration
features in the
box, with no
additional
licensing
costs
Capability
Hyper-V
(2012)
vSphere
Hypervisor
vSphere 5.1
Enterprise Plus
VM Live Migration Yes No1 Yes2
1GB Simultaneous Live Migrations Unlimited3 N/A 4
10GB Simultaneous Live Migrations Unlimited3 N/A 8
Live Storage Migration Yes No4 Yes5
Shared Nothing Live Migration Yes No Yes5
Network Virtualization Yes No VXLAN6
1 Live Migration (vMotion) is unavailable in the vSphere Hypervisor – vSphere 5.1 required
2 Live Migration (vMotion) and Shared Nothing Live Migration (Enhanced vMotion) is available in Essentials Plus &
higher editions of vSphere 5.1
3 Within the technical capabilities of the networking hardware
4 Live Storage Migration (Storage vMotion) is unavailable in the vSphere Hypervisor
5 Live Storage Migration (Storage vMotion) is available in Standard, Enterprise & Enterprise Plus editions of vSphere
5.1
6 VXLAN is a feature of the vCloud Networking & Security Product, which is available at additional cost to vSphere 5.1.
In addition, it requires the vSphere Distributed Switch, only available in vSphere 5.1 Enterprise Plus.
vSphere Hypervisor / vSphere 5.x Ent+ Information: http://www.vmware.com/products/vsphere/buy/editions_comparison.html, http://
www.vmware.com/files/pdf/products/vcns/vCloud-Networking-and-Security-Overview-Whitepaper.pdf http://
www.vmware.com/products/datacenter-virtualization/vcloud-network-security/features.html#vxlan
30. Failover Clustering
Unmatched
Scale
Failover
Clusters
support 64
Nodes and
8,000 VMs
Flexible VM
Clustering
iSCSI, Virtual
Fibre
Channel &
SMB 3.0
clustering
Highly Secured
Cluster Storage
BitLocker Drive
Encryption for
improved
security of
traditional and
Enhanced
CSV 2.0
Deeper
integration
with
storage
arrays for
31. Failover Clustering
3 Levels of
Availability
Host, Guest
OS &
Application
Level
Protection
Cluster-
Aware
Updating
Eliminate
downtime
associated
with cluster
Failover
Prioritization
Controls the
order in
which VMs
fail over or
start
VM Affinity
Rules
Ensure VMs
stay together,
or apart
within the
cluster
33. VMware Comparison
Only Hyper-V
provides
Guest OS
Application
Monitoring in
the box, with
no
additional,
expensive
add-ons
Capability
Hyper-V
(2012)
vSphere
Hypervisor
vSphere 5.1
Enterprise Plus
Incremental Backups Yes No Yes1
VM Replication Yes No Yes2
NIC Teaming Yes Yes Yes
Integrated High Availability Yes No3 Yes4
Guest OS Application Monitoring Yes N/A No5
Failover Prioritization Yes N/A Yes6
Affinity & Anti-Affinity Rules Yes N/A Yes6
Cluster-Aware Updating Yes N/A Yes6
1 VMware Data Protection is available in Essentials Plus and higher vSphere 5.1 editions
2 vSphere Replication is available in Essentials Plus and higher vSphere 5.1 editions
3 vSphere Hypervisor has no high availability features built in – vSphere 5.1 is required.
4 VMware HA is built in to Essentials Plus and higher vSphere 5.1 editions
5 VMware have made APIs publicly available, but actual application monitoring is not included
6 Features available in all editions that have High Availability enabled.
vSphere Hypervisor / vSphere 5.x Ent+ Information: http://www.vmware.com/products/vsphere/buy/editions_comparison.html and
http://www.yellow-bricks.com/2011/08/11/vsphere-5-0-ha-application-monitoring-intro/
34. VMware Comparison
Hyper-V
provides the
most
flexible
options for
guest-
clustering,
without
sacrificing
agility &
density
Capability
Hyper-V
(2012)
vSphere
Hypervisor
vSphere 5.1
Enterprise Plus
Nodes per Cluster 64 N/A1 32
VMs per Cluster 8,000 N/A1 4,000
Max Size Guest Cluster (iSCSI) 64 Nodes 16 Nodes2 16 Nodes2
Max Size Guest Cluster (Fiber) 64 Nodes 5 Nodes 5 Nodes
Max Size Guest Cluster (File Based) 64 Nodes 0 Nodes3 0 Nodes3
Guest Clustering with Live Migration Support Yes N/A1 No4
Guest Clustering with Dynamic Memory Support Yes No5 No5
1 High Availability/vMotion/Clustering is unavailable in the standalone vSphere Hypervisor
2 Guest Clusters can be created on vSphere 5.1 using the in-guest iSCSI initiator to connect to the SAN, the same as
would be configured in a physical cluster. Support of guest operating systems up to Windows Server 2008 R2
means 16 node clusters are the maximum size on vSphere 5.1
3 VMware does not support VM Guest Clustering using File Based Storage i.e. NFS
4 VMware does not support vMotion and Storage vMotion of a VM that is part of a Guest Cluster
5 VMware does not support the use of Memory Overcommit with a VM that is part of a Guest Cluster
vSphere Hypervisor / vSphere 5.x Ent+ Information: http://www.vmware.com/pdf/vsphere5/r51/vsphere-51-configuration-maximums.pdf, http://
pubs.vmware.com/vsphere-50/topic/com.vmware.ICbase/PDF/vsphere-esxi-vcenter-server-50-mscs-guide.pdf, http://kb.vmware.com/kb/1037959
36. Summary
Hyper-V: A More Complete Virtualization Platform
Scalability,
Performance &
Density
Secure
Multitenancy
Flexible
Infrastructure
High Availability
& Resiliency
37. Deep technical content and
free product evaluations
Expert-led, no cost, hands-on
technical training events
Free, online,
technical courses
Download Microsoft software
trials today.
Find an IT Camp near you. Take a free online course.
Technet.microsoft.com/evalcenter Technet.microsoft.com/globalitcamps microsoftvirtualacademy.com
At the TechNet Evaluation Center you can
download free, trial versions of Microsoft
software, with no feature limits. Dozens of trials
are available – all at no cost.
Try Windows Server 2012 for up to 180 days.
Download the Windows 8 Enterprise 90-day
evaluation. Or try Windows Azure at no-cost for
up to 90 days.
Microsoft IT Camps are no cost, hands-on
technical training events for IT professionals led
by Microsoft experts, centered on the issues
and workloads you’re tackling in your
environment today.
New IT Camps cover topics including Windows
Server 2012, Windows 8 for IT professionals,
Windows Azure and more.
Microsoft Virtual Academy provides free online
training on the IT scenarios that are important
to your company and your career.
Learn at your own pace and boost your IT skills
with over 100 courses across more than 15
Microsoft technologies including Windows
Server, Windows 8, Windows Azure, Office 365,
virtualization, Windows Phone, and more.
Now we’ve talked about a number of those capabilities there but I think it’s important to take a quick look at some of those capabilities in action and show how they’re set up and configured, so without further ado let’s duck into a demo.
Okay, so here we are in our demo infrastructure where we’ve got four physical Hyper-V hosts all running Microsoft Hyper-V Server 2012 and the first three here, so node 2,3 and 4 are configured in a cluster. Now all three of these hosts in a cluster are also configured to support fiber channel and if we look at node 4 specifically we go over to our virtual SAN manager here, now Hyper-V allows you to define virtual SANs on the host to accommodate scenarios where a single Hyper-V host is connected to different SANs through multiple fiber channel ports. And a virtual SAN defines a named group of physical fiber channel ports from the host that are connected to the same physical SAN. For example assume that a Hyper-V host is connected to two SANs, production and a test SAN. The host is connected to each SAN through two physical fiber channel ports so in this example you might configure two virtual SANs, one named production, that has two physical fiber channel ports connected and one named Test SAN, which has the two remaining physical fiber channel ports connected. Now you can configure as many as four fiber channel adaptors to a virtual machine and associate each one with a virtual SAN, so that means a single VM could be actually attached to four SANs. So when we connect, when we create a virtual fiber channel SAN we’re literally just ticking the boxes to represent the physical HBAs in the host that are going to allow our virtual machines to ultimately connect through to that SAN. So we’ll cancel out of there and we’ll look at a virtual machine settings that we’ve already configured here.
So I’ll look at the settings of this virtual machine, and we’ll see that this particular VM already has two fiber channel adaptors here and if you want to add more we’ll use the add hardware but I can’t add a fiber channel adaptor while a VM is running it’s something I need to do when it’s off. In this particular virtual machine what you’ll notice is in this particular fiber channel adaptor is that it’s got two sets of addresses; address set A, and address set B. now to support live migration of VMs across Hyper-V hosts while maintaining fiber channel connectivity two worldwide names are configured for each fiber channel adaptor, set A and set B. And Hyper-V automatically alternates between the set A and set V worldwide node addresses during a live migration. So this ensures that all nodes are available on the destination host before the migration and that no downtime occurs during the migration. But when you’re masking and zoning your LUNs on your fiber channel storage you take both of these addresses into account and that’s the address set for this particular fiber channel adaptor and the other one. But what does this look like inside the virtual machine? Well let’s take a quick look
So I’m going to double click on demo 6 and go into that virtual machine, have that connection open already and if we look inside Device Manager of this virtual machine you’ll see down here on the storage controls Hyper-V fiber channel HBA, the representation of that HBA but also you’ll see here that because I’ve got two fiber channel adaptors I’ve actually enabled MP IO inside the virtual machine or Multi-pathing, I’ve done that through the same configuration wizard as I would do if this was a physical host so I’ve got that consistency. And what we’ve seen by importing the relevant HP drivers, I’ve been able to see the actual HP Multi-path device that is in my environment. So consistency, a consistent user experience.
Now, let me come out of there and let me show you this particular disk and you’ll see that it’s a fiber channel disk, it’s 50 GB in size and we’re consuming around about 4 GB maybe just less, on this drive, but if we double down into that drive you’ll see that actually on this drive I’ve got about 126 GB worth of content or in this case virtual hard disk files. So I’ve got 126 gigs worth of stuff on the drive yet I’m only consuming around about 3,4 GB. What’s going on there?
Well one of the new storage features within Windows Server 2012 is an inbox deduplication, so if we go to the local server and I look at file and storage services and volumes, what you’ll see is for this fiber channel disk which I’ve turned on deduplication I‘m at a deduplication rate of 97% so we’ve deduplicated more than what a single virtual hard disk file would consume on it’s own, we’re down to around 3 to 4 GBs, that’s a huge savings, saving 123 gig versus what we would need to have in place if we didn’t have deduplication. That’s phenomenal saving, that’s a great example of using technology that’s built into Windows Server 2012 to help us be more efficient and effective.
Now, what you’ll also find is that we have a new capability within Windows Server 2012 that I highlighted in the slides earlier on called storage spaces. And if I go to the storage pools section here you’ll see that I’ve got actually a storage space not created this is just a default to show me my physical disks down here and what you’ll notice is this particular machine has got three physical disks attached to it, each with 64 TB in size, that’s huge, that’s a huge amount of storage space that I’ve got available for each virtual machine. But these are just regular disks at this point, they’re not a storage space, so I want to go to create a new storage space or pool, we’ll go next, I’m going to give it a name so we’ll call it space 1 for now and it’s going to look at the disks that are available in the primordial pool there to use as part of this space. So we’ll go next, we’ll select these physical disks and notice I can change the allocation as well do I want a particular disk to be a hot spare or part of the pool automatically, we’ll go next and you’ll see I’m going to have 192 TB of capacity which is great, we’ll go next, and we’ll go create. It’s updating, it’s finishing the process and that’s great and notice I’ve got this opportunity to create a virtual disk when this wizard closes. So we’ll go ahead and we’ll create that virtual disk, so we’ll go close, so now I’m creating a virtual disk, from a storage pool and this virtual disk could be used by something else, now it’s important to not think of necessarily a virtual disk like a virtual hard disk, this is a virtual disk that could have a volume on it that could be used for a file share or some other workload. So we’ll go next, we’ll choose the space that I’ve created, Space 1, we’ll go next, we’ll give this a name, Disk 1, we’ll go next, now do I want this to be simple, so data’s going to be striped across all three of these disks, giving me maximum capacity and throughput but there’s no resiliency. And mirror would enable me to duplicate data across two or three physical disks increasing reliability but I might sacrifice a little bit of capacity. However parity is striped across physical disks, increasing reliability, though somewhat reducing capacity, I’m going to go with parity on this one. Now, do I want to provision this data? Fixed so use the complete space straight away, or thin provision, I’m going to go thin provision, go next, size, I’m going to choose a virtual disk size of let’s say 100 TB, then we’ll go next, then we’ll go create. So it’s creating all of those elements and now the final step of the wizard is creating a volume, because what we’ve got at the moment presents to Windows as a blank 100 TB disk now that’s part of a pool of 192 TB maximum capacity, we’ve not got a volume which we can actually store data on, so we’ll go close and we’ll make sure that we’ve got that tick box ticked, we’ll go next, which particular machine am I going to provision this to on which disk, so you’ll see it’s 100 TB this virtual disk that we just created is Disk 1, the disk I’m creating is going to be called Disk 5. go next, what volume size, let’s make it the full amount, drive letter, yes let’s leave it as the default, file system, volume label, HugeDrive, NTFS, all that can be left as default, do you want to enable dedups as part of the wizard that is engrained within Windows Server, we’ll go next and we’ll go create. So very quickly we’ve taken three effectively blank disks, we’ve transformed them into a storage space, we’ve provisioned a virtual disk on that storage space and from there we’ve now created a volume of 100TB, do we actually have physically 100TB in my infrastructure to use? No, but this has all been thin provisioned, I’m going to close that out and you’ll see now that we’ve got our storage space and you can see that we’ve got a thin disk provisioned and it’s currently allocated 2GB of the 192 maximum and that’s been used, if we look at the volume view here, there’s our F drive, hugedrive 100 TB, it knows it’s thin provisioned so it’s aware of this kind of stuff. And if I go to Windows now, there’s our huge drive, 100TB in space, it’s consuming minimal amounts at this point in time but as we copy data to it, we’ll see more.
But what can we do to expand the infrastructure, what if we want to add another 64 TB disk? Well, if I want to do that I can minimize this particular VM, I’m going to go over to cluster manager because this virtual machine is a clustered machine I should be doing everything really from here from an administrator perspective, I’ll go to settings of the VM and you’ll see it’s already got 3 64 TB disks attached to this VM those are the three disks we saw earlier but I’m going to go to SCSI controller, hard drive, add, remember this virtual machine is running at this point in time, we’ll call it Disk 4, in fact we’ll go no, next, VHDX, Dynamic, so I’ll make it small to begin with and grow, we’ll call this Disk 4, it’s a VHDX file which allows us to have that increased capacity I’m going to choose where I’m going to store this, so I’m going to store this with my other virtual disks in that folder there, I’ll go next, what size, 65 or 64 TB and we’ll go finish. We’ll go apply, and if I go back to that virtual machine now and we do a scan for disks, there, we found our remaining 64 TB disk that we’ve just created, what can I do with that? Well, I can go to my storage pools I can select space 1, I can add an extra physical disk to it, tick, okay, and just like that we’ve increased the capacity of our storage pool to 256 TB and if I scroll down to our virtual disk I could now go ahead and extend this virtual disk to something even greater and then subsequently increase the size of the volume as well.
You’ll notice I could also if I wanted to remove a disk. So we’ll go yep, we’re going to rebuild everything on the fly, we’ll go yes, ok, so very quickly we’ve added a disk to a storage space, we’ve removed a disk from a storage space, incredibly easy, simple and effective.
<number>
04/06/13
And it’s important let’s have a look at some of those things through a quick demonstration. So here we are back in the demo environment, this is slightly different one to show some different demonstrations and capabilities. First of all I want to talk about the improvements in Dynamic Memory which as we discussed before help organizations to realize better density and a better return on investment. so what we can see here is one of my VMs here Demo01 has been enabled with Dynamic Memory, it’s got a minimum and maximum between half and one GB the current assigned memory that Hyper-V has given our VM is half a GB the minimum but the demand inside the guest OS is relatively low, it’s actually lower than the minimum. But you’ll notice this new feature, this option we’ve got within Windows Server 2012 and Hyper-V is to specify the startup memory. Because certain VMs start up with more memory to get certain things done initially, certain things started and then once they’re calmed down they can then drop their memory requirement below that startup memory, down towards what the minimum is. We’ve kept them as the same for this example but it’s up to you. What we’ve got here though is the opportunity to now start to hot add dynamic memory to really make it dynamic and scale up that workload as it grows.
So by going into this particular virtual machine here which we’ve got open here, COM Demo1 and we can run a little application that’s going to consume memory, so it’s going to increase the memory usage of our virtual machine, so if we minimize back out of there we’ll see warning, memory demand is higher than what the maximum is here so maximum 1GB, memory demand is higher, so I can go into the settings of this VM now, this is new in 2012, go to memory and I can adjust the maximum on the fly to something more appropriate, so I’ll choose 2 GB, I’ll apply, I click ok, and you’ll see now the assigned memory, Hyper-V starting to give it a little bit more, but it’s continuing to grow, the demand that little application that’ s running inside there is continuing to need more and more memory, so we could go ahead and assign more but in fact, I’m just going to kill it. And memory demand what you’ll find now will start to shrink down, so the memory demand has dropped considerably and now the assigned memory, Hyper-V will start to reclaim that even though there’s no contention, over time it will start to claim that memory back and give it other virtual machine when they need it.
Now we mentioned before resource metering is a way to track the certain metrics associated with virtual machines even if they migrate and float around the infrastructure using some of the migration technologies.
So what I want to do is just show you how easy it is to set that up, now in my notepad file here I’ve got an example of a script that you would write in PowerShell to set this up. So at first on the host you would enable resource metering so a simple script there, then you would set your interval, so how often you’re going to query and check and then finally this final script here enables those, I’m just going to copy and bring up PowerShell, this enables us to actually query and product what’s been used by the virtual machine. And you’ll see a number of the virtual machines that have been running that period of time, the average CPU and megahertz, the average memory, the maximum memory and we’re able to use that particularly in solution search of System Center to build show and charge back models to represent the usage and give the information back to the business. So really easy and obviously with this in PowerShell this can be piped out to other systems as well.
So that was a little bit about resource metering and I want to finish on quality of service and show you how easy it is to define quality of service for certain key workloads and applications. Now to do that currently if I look inside this particular virtual machine demo001, I’m going to run this little client network application, it’s going to fire towards a receiver if you will where I’ve already set up a certain amount of traffic in a certain amount of time. and you’ll see in this case in five seconds it was able to send 230 MB of data at a rate of 385 M/sec. Now if I press continue there and close that and I come out of this particular virtual machine and look at the settings on the host of that VM, in the network adaptor here that’s connected and we turn on bandwidth management and we set a maximum of say 50 Mbps so we’re stopping this virtual machine being particularly noisy which could impact other virtual machines. If you’re interested in giving a certain VM a certain amount of SLA so a minimum level of bandwidth available you can select and enter information in this minimum column as well. But I’ll go ok there, I’ll go back to the virtual machine and we’ll run the same test again and you’ll see this time it didn’t take any longer because it was five seconds an interval but you’ll see the transfer was a lot smaller at a lot lower transfer rate that we were controlling through quality of service. It’s a really simple demonstration but very, very powerful stuff available to customers of all shapes and sizes, very, very useful indeed. And with that, we’ll go back to the slides.
<number>
04/06/13
And I want to take a moment to show you these capabilities in action so let’s take a look.
So here we are again in my demo environment and I’m going to bring up failover cluster here and you can see that I’ve selected a number of machines that are currently running on our cluster it’s a three node cluster remember, all three running Hyper-V Server 2012 and these four virtual machines that I’ve selected I’m going to right click, move, live migrate, and now notice what’s new in Windows Server 2012 it’s almost this guidance of where we’re going to distribute the virtual machines to. You’ll notice all the virtual machines are currently on node 4 and if I selected best possible node Hyper-V and the clustering management tools are going to move the virtual machines and distribute them a little more evenly. But I’m going to choose to select to move them all to a single node and I’m going to choose node number 2, and we’ll go ok. And you’ll see that all of those virtual machines are migrating simultaneously, as quickly as possible utilizing that network bandwidth as much as possible, final virtual machine going across there, notice this final virtual machine is actually one that’s got our virtual fiber channel adaptors attached to it as well, so that should have been moved across.
You’ll notice that one’s got a little bit more memory than the others and therefore it took a little bit longer but that was a very easy and simple demonstration showcasing multiple, simultaneous virtual machine live migration.
What I want to do now is select a slightly different VM Demo number 5 and I’m going to choose move again but this time I’m going to choose virtual machine storage, so I’m going to do a live storage migration. And we’ll click virtual machine storage, notice I’ve got all of the things that make up a virtual machine so the disk, snapshot, configuration files and so on, and I want to choose where I want to drop them. Currently it’s located on volume 1, but I’m going to choose to just click and drag and move to volume 3. so we’ll go start, and off it goes so we’re starting that virtual machine storage migration and that’s going to take a little time to complete, it’s a decent size due to the amount of storage there to move and if I go over to Hyper-V Manager here, we can see as it stands on #2 there migrating storage, 17, 18% and it’s synchronizing, it’s shifting those virtual machine files around to a different location, in this case a different LUN on the same SAN. Now, one thing I also want to show, the final area of migration is the shared nothing live migration. Because what we’ve shown initially is how we can move a multiple virtual machine, multiple running virtual machines between cluster nodes, we’ve also shown the movement of the disks all of this is without interruption for those running virtual machines but now I want to choose to move a virtual machine between two physical servers but share nothing but an ethernet cable.
Now over on node 3 here I’ve got a virtual machine and I’m going to right click, I’m going to click move, and notice I’ve got this move wizard here that’s going to ask my okay, what do you want to do, just move the disk or move the whole VM and it’s disks, everything about it? We’ll go with the whole VM, we’ll go next. What’s the destination, so I’m going to choose CON demo5 for this particular one and we’ll go next. The move options, do you want to move everything to one place, do you want to separate things out so move the one virtual disk to here, move the config file to here, I can separate all of that out or just move the running state of the VM but the virtual machine disks must be currently on shared storage, that’s not the case so we’re just going to move the top one. We’ll go next, I’ll choose a folder to place it on the target machine I’m just going to go there, that’s fine, the D drive and then we’ll go next and finish and that’s going to start and that's performing the move and that’s going to take a minute or two but don’t forget this is moving the running state and at the same time the virtual disks. So we’re literally picking up the whole of that machine and transporting it to another physical server. If you’ve been following along, nodes 2,3 and 4 is a cluster and Hyper-V 5 is a stand alone server. So we’re actually moving a VM in and out of a cluster in this particular example we’re moving it out of the cluster. So that’s the flexibility that Hyper-V in Share Nothing migration brings, moving VMs between stand alone servers, stand alone to cluster, cluster to cluster, cluster back to stand alone, complete flexibility utilizing Shared Nothing Live migration. And you’ll see it’s zooming through, zooming along, we’re just finishing wrapping up the process now and there, done, disappeared from node 3 and you’ll see it’s now here on node number 5 and you’ll see it’s still running there, you can make out in the little screen there. great stuff, great flexibility, agility and all in the box with Windows Server 2012 and Hyper-V.
<number>
04/06/13
So let’s have a look at some of this stuff in action, now in this first demonstration I want to take a quick look at that feature we called Hyper-V replica earlier on and to enable that I’m going to failover cluster manager here and the reason I’m going to failover cluster manager is because the virtual machines that I want to protect with Hyper-V replica are running on the cluster. So I need to enable this features on this VM from failover cluster. What you’ll also notice on this cluster is I’ve got a little service or role running here called the Hyper-V replica broker and that’s an important service that you need to enable on the cluster if you are going to be replicating in and out or in or out of the cluster, so very simple to enable, all the documentation’s on TechNet, but it’s just a role that you turn on, on the cluster. But now that it’s turned on I can start to enable Hyper-V replica for key virtual machines. And to do that I’m going to select this particular virtual machine VM4 and I’m going to right click, replication, and enable replication and this brings me up the Hyper-V replication wizard and what I can do is firstly specify where I’m going to replicate this to and I’m going to choose to replicate to a server outside of the cluster, so it could be a server that’s outside my environment, another datacenter, you decide. And I can choose depending on where I’m going to replicate that to whether I want to secure it with a certificate, what port I want to send this information over and whether it’s going to be compressed as it goes over the network. But I’m going to go next, I can choose which virtual disks are going to be replicated and in the example here is for example a VHD used for a dedicated paging file, you may not want to replicate that is it unnecessary, in which case you could unselect that one. But I’m just going to choose these particular VHDX file. We’ll go next, how many recovery points do I want to deliver to this virtual machine, so I’m going to say that we’ll keep only the latest recovery point, but I could add more and I could also turn on the VSS capability for application consistent snapshots or copies, so that gives me that extra level of resiliency but I do need additional storage to do that. But I’m going to go only the latest for the time being, we’ll go next, now I’m going to replicate 12 GB of data and how do I want to get that there initially? Well I could just send that over the network, I could send an initial copy using a USB disk, or some other form of external media, I could choose something I’ve already got on that site like back up that’s been restored that I could then replicate onto and I can choose when I’m going to start this replication. So I’m going to choose right now, immediately over the network and we’ll go next. And we’ll go finish and off that goes, that started replicating now and you can see here, initial replica in progress, sending initial replica and what you’ll notice is if I go over to Hyper-V Manager because that’s where node number 5 is you’ll see receiving changes here, replicating across the network from A to B in this case hosting the cluster out of the cluster. And while it’s doing that, because that's going to take a few minutes, let’s look at what the situation looks like from an actual failover perspective.
So we’ll go back to failover cluster manager and we’ll select demo 3 which I’ve already set replication up with and I’m choosing now to instigate that failover from A to B. but firstly because I want to make sure that this is a clean migration one of the things that a planned failover I know that I’m going to need to failover in the event of a disaster here it needs the virtual machine to be shut down or at least turned off, it doesn’t have to be shut down cleanly but you may want to do that. I’m going to do that because I’ve got that opportunity, I’ve been warned of an impending natural disaster on the way so I’ll shut that VM down and give it that time to shut down cleanly, I’ll then right click, replication, planned failover and that’s going to run through a wizard and you’ll see that the first prerequisite check is to check that the VM is turned off. So whether you shut it down cleanly or switched off the virtual machine, as long as it’s in the off state it’s going to be able to continue. And check configuration for allowing reverse replication so is it going to be able to replicate back this way when we’re doing this planned failover, so we’ll go failover, and I’ll jump over to Hyper-V manager quickly and you’ll see it was just about starting that virtual machine now, so it’s off on the cluster and now they’re starting it up in this environment and this is flashing away to say everything was fine, completed successfully. So you’ll see here now we’ve started up and it’s up and running and it sent those final changes that we perhaps hadn’t replicated prior to shutting down and it’s also replicating back the other way, that’s really useful as well.
Let’s just look quickly at some of the settings inside this VM concerned with replica. Because if I look at the settings for this VM you’ll see I’ve got this little option here called failover TCP/IP so these are the settings that I could apply but when the VM fails over to the other sites these IP addresses are injected and used for that virtual machine once it’s failed over. So that’s pretty useful for it saves me when configuring those kind of settings and IP address in a panic when we’ve actually gone into a full failover situation, so I’ll cancel that out.
Now, if I go back to the cluster this one is off, now if we had to of lost the other site, so lost the primary this now being the secondary, I have the option to do a failover which basically says okay, you’ve lost the main sites, we’ve lost the primary, bring up the secondary. Simple as that. Test failover however let’s me spin up and bring these virtual machines up and running in an isolated type environment which means that I can do that while the primary workload is still running. 7:28
<number>
04/06/13
But I think it’s important to take a look at some of these technology and features in action so that’s what we’ll do in this demo. Now in this final demonstration I want to take a look at some of the key resiliency capabilities that are new in Windows Server 2012. And the first one is around this ability to offer extra levels of availability to your key applications and services. Now if I go into our cluster manager here and have a look at Demo 06, it looks like any normal virtual machine but if I click summary here, what you’ll notice is the monitored services listed here is print spooler, now what’s that about? Well this VM has been enabled for the monitoring services that third level of availability so if I click more actions, configure monitoring I’ve ticked the box for print spooler that are other services that I could protect and other things that I may have installed inside the virtual machine would appear here but you have to meet the pre-requesites obviously. I’m going to cancel out of there because I’ve already enabled that and what that means is if that print spooler has a problem, and I’m going to connect to that virtual machine now so we can have a look inside, and we’ll close the storage space dialogue box that we had open before and I’m going to bring up services and we’ll see that there’s the print spooler currently running and if we look at the properties, recovery, first failure, restart the service, so if it stops, restart the service, second failure, restart the service, this is Windows stuff, this is what Windows Server does inside the guest operating system. But what I’m going to do is I’m going to run a little process here to kill the print spooler, sounds very harsh and very cruel, it’s for the purpose of demonstration so it should be fine, so I’m going to go kill and it’s been terminated. So if I refresh it’s gone, if I refresh again it’s back on again because Windows has kicked in and saved it. If I do it again, hit refresh, it’s gone down here, and it’s back. And if I do it again, it shouldn’t, it’s gone, keeping refreshing, hitting f5 and it’s not coming back, okay, so what’s happened on the cluster? Well, let’s minimize out of there and let’s go back to the cluster and this will probably take a few seconds to pick up but what we’ve got now within the clustered environment, if I make this a little bit bigger, the cluster is monitoring this specific service, the Hyper-V host is monitoring this specific service and when it polls and when it checks to see what’s happened or see if this service is still running, it’s going to encounter an error and it’s going to put an event in the event log at the cluster level. So your monitoring tool, your management tool is looking for that particular event can respond to that. Now I say your monitoring tool or management tool can respond to that, I can actually choose to respond from within a cluster, now if I look at resources very briefly in fact we just saw it flick over there, running the application in VM critical, you’ll see there’s been a problem noted, if I look at show critical events and it’s been raised at the cluster level as well, if I show critical events we can see that today’s date, 3:38 cluster resource on clustered role received a critical state notification, indicates a application or service inside the VM in unhealthy and that’s the event ID, 1250. Now, what I could have done if I click on resources down here, and right click and look at properties, in the settings view here, I’ve unticked this little box here, so heartbeat settings, yes, enable heartbeat monitoring for the VM, that’s fine, but enable automatic recovery for application health monitoring if I’d have had that ticked as soon as that error was detected, the Hyper-V and the cluster would have restarted the virtual machine and if that didn’t fix the problem it would have also then restarted the virtual machine, another alternative host, but I chose to untick that box to show you what the event and the problem would look like. But that’s what the kind of level and automation and remediation we’re building into the product but if I’d have ticked that box it would have restarted that VM.
But you might think well, I might not want to restart that VM because there might be something else relying or maybe another dependency somewhere else and that’s why you bubble up that event, that critical event that was registered into your more comprehensive management tools, tools like System Center Operations Manager is a good example.
Now, the final demo I want to give and want to show you is still within this particular infrastructure but it’s a feature known as cluster aware updating and cluster aware updating is a new feature in 2012 that enables me to remediate complete cluster infrastructures, moving workloads around that cluster no matter what’s on the cluster, whether it’s file service, Hyper-V VMs, DHCP services, whatever it may be. Now notice that this was last run at 3am and it succeeded, now I wasn’t up at 3am to do this, I had it automatically set up to automatically remediate my cluster. Now, how I do that, I configure my cluster self updating features, now I can do this manually if I want to, but sometimes having it automatic is pretty beneficial. I’ll go next, enable self updating mode, if I want to do it manually I can disable it, but whenever I want to update the cluster as a whole I would come to this UI and do it manually. But I’m going to go next, I’ve chosen monthly, I’ve just changed it from daily, go next, it’s going to use these options which I could tweak, so reboot time out, post scripts, etcetera, we’ll go next, which updates, next and apply. And that’s done, now what’s important here is your cluster nodes are still going to pull from the essential patching source, so if that’s W source, if that’s Configuration Manager they’re still going to use that, all this is doing is triggering a polling of what updates are available for each individual node and then it’s orchestrating the delivery of those patches, the movement of the workloads off the cluster nodes and then the necessary reboot if necessary of that host and then moving the workloads back. And you’ll see if I was to preview updates for this cluster because I’ve already run it multiple, every night for the last couple of weeks there’s probably not going to be any, no, no updates found. But that’s what I would do if there were updates found I could check them, see which ones are applicable, we’ll go close and then I would click apply and it would orchestrate the whole process of deploying the patch to that particular cluster node, moving the VM workloads and so on.
But if I do want to see what’s been happening in the past I can generate a report so I can see what’s happening in the last month or so or in this case I could put whatever start or end date I’d like but I’ll just choose the last month and you’ll see for most nights zero, zero, zero, nope there was a couple of updates pushed out on the 28th of November we can see what they were for specifically, information about that, great. There were a few more a bit later on, so December there was 27 so 9 a piece and then finally just a day or so ago security update there as well. So providing these has been approved in W source so each cluster node can actually see them because when the cluster that we’re updating is triggering that node to scan what updates are available for it, if they’ve not been approved in W source or Configuration Manager whatever your patching tool is then the cluster nodes aren’t going to be able to see it and therefore it’s not going to appear in this list and it’s not going to be pushed down from W source.
So that’s a little bit about cluster aware updating it orchestrates that whole process of virtual or physical cluster nodes been rebooted, moved around, shifted all of that kind of stuff, extremely powerful and great for businesses of all shapes and sizes and with that we’ll go back to the slides.
<number>
04/06/13