\\\\ Best Practices for Managing Storage with Hyper-V Learn how storage functionality changes with the newest Hyper-V release and how it differs from VMware. Get details on specific Hyper-V features including Offloaded Data Transfer, Virtual Fibre Channel and Hyper-V Replica. By brien m. posey Expert Briefing
SBest Practices for Managing Storage with Hyper-V torage professionals waiting on Hyper-V 3.0, due for release with Windows Server 8, can use this Expert Briefing to guide their way through this critical product. Microsoft expert and Storage Media Group contributor Brien M. Posey answers key questions from the SearchStorageVirtualStorage. com editorial team about supporting and managing storage for Hyper-V and explains the most relevant technical features within the product. How does supporting storage for Hyper-V differ from supporting storage for VMware? Microsoft s Hyper-V and VMware Inc. s vsphere have many similarities but also several differences when it comes to supported storage. As a general rule, Hyper-V is much more flexible with regard to the types of storage that it supports, because it s built into the Windows operating system; vsphere, meanwhile, has a fuller range of enterprise-level storage management functionality, though Microsoft is making strides to level that playing field with the upcoming version of Hyper-V, Version 3.0 Drilling down into that a bit, Hyper-V is able to store virtual hard drives and other virtual server components on just about any storage that is accessible to Windows. This means that the files can reside locally on SATA, SAS or solid-state storage, or they can reside on remote storage that is connected using Fibre Channel or iscsi. VMware also supports local and remote storage, but it does not allow you to store virtual machines within a random folder on an NTFS partition the way that Hyper-V does. Instead, VMware requires you to create a datastore to accommodate virtual machines. It is also worth noting that VMware has very limited support for SATA drives, so you will almost always have to use either SAS drives or remotely connected storage. What are the main storage-related changes expected in Hyper-V with the release of Windows Server 8? Hyper-V 3.0, which is expected to be released this fall with Windows Server 8, will feature a number of storagerelated improvements. One such improvement is that Windows Server 8 will support the SMB 2.2 protocol. What this means for storage is that because the protocol is more efficient than previous versions of SMB, it will be possible to store virtual machines on file servers rather than having to use directexpert briefing may 2012 2
With Shared Nothing Live Migration, it will be possible to perform live migrations of virtual machines without shared storage. attached storage or Fibre Channel/ iscsi-attached storage. Another new Windows Server 8 feature is file system deduplication. This feature has tremendous potential for Hyper-V. There is usually a lot of redundancy among virtual machines. For instance, each virtual machine might have the same operating system files. Deduplication will allow virtual machines to consume far less physical storage space. This can lead to higher consolidation rates and lower storage costs, and in some cases it will make using solid-state storage affordable. Another new feature is something called Live Storage Migration. The current version of Hyper-V has the ability to move a virtual machine s storage between volumes on a host server using a process called Quick Storage Migration. The problem is that doing so causes a service disruption. Live Storage Migration will allow virtual machine storage to be moved without interruption. Finally, there s Shared Nothing Live Migration, which is the most exciting of all of the new Hyper-V 3.0 features, which will eliminate the need for shared storage in Hyper-V. Small and midsized companies will benefit the most from this change, while larger organizations will probably continue to use shared storage for performance reasons. How does Shared Nothing Live Migration change things for data centers running Hyper-V? Prior to Hyper-V 3.0, live migration was possible only within a clustered Hyper- V deployment. Furthermore, the cluster had to be built around a shared storage architecture. This shared storage was connected to each host in the cluster by either an iscsi or a Fibre Channel connection. The requirement for shared storage put Hyper-V clustering out of reach for many smaller shops. Shared storage tends to be expensive, and it has to be designed in a way that delivers adequate performance for all of the virtual machines. Sometimes this can be a tall order. With Shared Nothing Live Migration, it will be possible to perform live migrations of virtual machines without shared storage. Not only will the need for shared storage be eliminated, clustering will also not be needed. expert briefing may 2012 3
What is Hyper-V Offloaded Data Transfer, and how does it work? Offloaded Data Transfer [ODX] is a new feature that is designed to decrease the amount of time that it takes to read from or write to storage. Currently, processes such as merging snapshots, moving virtual hard disks or compacting virtual hard disks are time-consuming because the operating system has to be involved in the process. ODX will greatly increase the efficiency of read and write operations by handing the operation off to the storage hardware so that the operation can be performed natively at the storage level rather than requiring the operating system to orchestrate the entire process. ODX is similar to the functionality delivered by VMware s vstorage APIs for Array Integration [VAAI], though there are differences between the two. To take advantage of ODX, the virtual machines must be hosted on ODXcapable hardware. The best part of the Offloaded Data Transfer feature is that with ODX-capable hardware, the functionality is enabled automatically, without intervention from an administrator. How does Storage Spaces work in Hyper-V 3.0? Storage Spaces [which has also been referred to by Microsoft as Storage Spaces and Pools ] is a new feature in Windows 8 and Windows Server 8 that enables thin provisioning. The idea is The best part of the Offloaded Data Transfer feature is that with ODX-capable hardware, the functionality is enabled automatically, without intervention from an administrator. to abstract a storage volume from the physical disk. The process works like this: An administrator defines a storage space of a specific size and then adds as much physical storage as he or she wants until the storage space s predefined size is reached. The physical storage that makes up a storage space is managed by the operating system and is totally transparent. In order to protect against disk failures, Windows 8 will offer a number of resiliency techniques. For example, two-way mirroring, three-way mirroring and parity are all supported. Thin provisioning will make it possible to create a large, mirrored storage space even when the underlying physical storage is not as large as the storage space that is being created. expert briefing may 2012 4
Is Live Storage Migration in Hyper-V essentially the same thing as Storage vmotion in VMware environments? How does it work? Yes, the two features are essentially the same thing. VMware s Storage vmotion allows you to migrate virtual machine disk [VMDK] files across storage arrays with no interruption of service. Hyper-V 3.0 s Live Storage Migration will accomplish the same thing. The functionality is relatively simple. When Live Storage Migration is initiated, disk reads and writes are still made against the source virtual hard disk. While these read and write operations are occurring, the virtual hard disk s contents are copied to the new destination as a background operation. Once the copy process is complete, write operations are mirrored to both the source and destination virtual hard disks. This allows any outstanding write operations to be completed. After the source and destination virtual hard disks are synchronized, the virtual machine is switched over to use the destination virtual hard disk, and the source virtual hard disk is deleted. What about Hyper-V s support for SMB 2.2; why is that significant? Prior to Hyper-V 3.0, the options for storing virtual machines have been somewhat limited. If a Hyper-V server is not a part of a cluster, virtual machines usually reside on direct-attached storage [although SAN storage is supported]. This could be anything from an internal storage array to individual physical hard disks. Clustered Hyper-V servers are required to use shared storage. Shared storage devices have to be connected through iscsi or Fibre Channel. While IT organizations often invest a lot of money in file server storage and in the infrastructure to back up those file servers, those servers cannot be used for storing virtual machines. This scenario changes with Hyper-V 3.0, which will gain support for the SMB protocol. That s because SMB Version 2.2 is more robust than prior versions with enough extra bandwidth and resiliency against network failures that, with the upgraded protocol, Microsoft will support storing virtual machines on file servers. [The extra bandwidth comes from SMB 2.2 s multichannel nature. If multiple paths exist between a source and a destination, the protocol can use those paths to increase the overall bandwidth.] How does Hyper-V Replica work, and why is it important? Hyper-V Replica is a disaster recovery tool that creates standby backup copies of virtual machines. The process of creating a replica involves the use of a primary host and a secondary host. When a replica is created, the VSS writer [a mechanism for creating application-consistent backups] creates a snapshot of the virtual machine on the expert briefing may 2012 5
primary host. The Replica function then copies the snapshot to the secondary host. That way, the organization has a standby server containing copies of its virtual machines, which are ready to be used at a moment s notice. Creating Hyper-V replicas does not require that the primary host and the secondary host have identical hardware, but the secondary host must meet some minimal standards. Specifically, the secondary server must have enough memory, storage space and processing power to host the replicated virtual machines. And, although not always an absolute requirement, to avoid compatibility problems, the secondary server should ideally contain the same class of processor as the primary host. What is Guest Fibre Channel in the new version of Hyper-V, and what benefits does it bring? Guest Fibre Channel [or Virtual Fibre Channel, as some Microsoft documents call it] is a new Hyper-V 3.0 feature that will allow virtual machines to directly connect to Fibre Channel-based storage. This new feature is important for a couple of reasons. First, Virtual Fibre Channel will make it possible to virtualize servers that require direct Fibre Channel connectivity. Previously, virtualizing such servers was not an option, unless users were willing to give up Fibre Channel connectivity and use pass-through storage. Guest Fibre Channel is also important because it will facilitate clustering of virtual servers over Fibre Channel. If such clusters already exist within physical servers, individual cluster nodes can be virtualized one at a time without destroying the cluster in the process and without having to take the clustered resource offline. n Brien M. Posey, MCSE, has received Microsoft s MVP award for Exchange Server, Windows Server and Internet Information Server. Posey has served as CIO for a nationwide chain of hospitals and has been responsible for the Department of Information Management at Fort Knox. He is a frequent contributor to TechTarget s Storage Media Group. expert briefing may 2012 6