What’s New in Windows 8 for Hyper-V Based Cloud Computing (Part 6) – Hyper-V High-Availability Features

If you would like to read the other parts in this article series please go to:

Windows Server 8 Continuous-Availability

In Windows Server 8, Microsoft shifts from a high-availability to continuous availability approach. Microsoft defines continuous availability as “software and hardware platforms designed to support transparent failover without data loss”. Windows Server 8 enhances existing software features (e.g., Failover Clustering, Live Migration, Storage Migration, and Guest Clustering) and adds new features (e.g., Hyper-V Replica) to improve continuous availability. Windows 8 Hyper-V also takes advantage of enhancements in storage, network, and server hardware designs to achieve continuous availability.

Failover Clustering

Windows Server 8 introduces massive scale improvements in the failover clustering features. While Windows Server 2008 R2 already supports large 16 node clusters, Windows Server 8 effectively quadruples the cluster size to 63 nodes capable of hosting 4000 VMs. This allows the creation of clusters with a high degree of flexibility, multiple maintenance nodes, and scaling ability adaptable across the range from small businesses to large multi-geographic enterprises.

With Windows Server 8, Cluster Shared Volumes (CSV) scale and compatibility are enhanced with integration as a core failover clustering feature. Enabling CSV requires only a single right click on a storage disk. Additionally, CSV natively supports storage filter drivers such as those used by anti-virus, backup, storage replication, and data protection products.

Live Migration

Live Migration, introduced in Windows Server 2008 R2, provides the ability to move a running virtual machine from one Hyper-V cluster node to another without experiencing any service interruption. New capabilities in Windows Server 8 Live Migration allow moving a running virtual machine from a Hyper-V server to any other Hyper-V server. In fact, you can use live migration to move virtual machines between nodes in a cluster, between nodes in different clusters, between a non-clustered Hyper-V server and a Hyper-V server node in a cluster, or between two non-clustered Hyper-V servers.

Live Migration in Windows Server 8 also supports multiple concurrent live migrations. There is no built-in limitation to the number of concurrent migrations; rather it is bounded by available resources such as the amount of available network bandwidth. In order to scale to a higher number of concurrent live migrations, you just have to provide additional resources, like network bandwidth, to the Hyper-V hosts between which you want to migrate virtual machines.

Live Storage Migration

System Center Virtual Machine Manager 2008 R2 introduced Quick Storage Migration to allow the migration of a virtual machine’s storage from one storage device to another with minimal downtime. Quick Storage Migration relies on Windows Server 2008 R2 Hyper-V and Background Intelligent Transfer Service (BITS) to perform the migration. Quick Storage Migration can migrate virtual disks of a running virtual machine regardless of storage protocols (iSCSI, FC) or storage type (internal, DAS, SAN).

Windows Server 8 provides Live Storage Migration as a native feature of Hyper-V. Live Storage Migration allows you to move virtual hard disks attached to a running virtual machine to a new location without incurring downtime. Common scenarios that require a live storage migration include upgrading storage devices, migrating storage between different levels of storage service, supporting backend storage maintenance, and redistribution of storage load. Live storage migration supports migrating virtual disks of a running virtual machine regardless of storage protocols (iSCSI, FC, SMB) or storage type (internal, DAS, SAN, file share).

At a high level, live storage migration performs the following operations during a migration:

  1. Make disk reads and writes to the source VHD
  2. Create a child differencing disk for new disk writes
  3. Copy the source VHD to the new storage destination
  4. Mirror new writes to the source and destination VHDs and finish replication of differencing disk from the source to the new storage destination
  5. Use the destination VHD for new reads and writes once the source and destination VHDs are synced
  6. Delete the source VHD

Network Interface Card Teaming

In Windows Server 2008 R2 and prior releases of Windows Server, if you wanted to team network interface cards (NIC) to prevent connectivity loss in the case of a NIC failure, you had to buy multiple NICs from a single vendor and leverage their software to enable NIC teaming. Even then, Microsoft did not support the solution and typically would require you to disable the NIC teaming when performing troubleshooting.

Windows Server 8 provides NIC teaming as a native operating system feature, and supports NIC teaming between different vendor adapters. For example, you can team Intel and Broadcom NICs together in failover or trunking modes using only the NIC teaming feature of Windows Server 8. This allows a more streamlined solution which does not rely on proprietary vendor technologies that have the potential to conflict with each other. In short, you get a fully supported NIC teaming solution from Microsoft. Of course, you still have the ability to implement another vendor NIC teaming solution if it offers special, required features that are not provided by Windows Server 8.

Three distinct NIC teaming modes are supported: switch independent, static teaming, and Link Aggregation Control Protocol (LACP). In order to leverage Static teaming or LACP, the network switch must support these modes. A single Windows Server 8 machine supports up to 32 NICs to team and you can manage NIC teaming using the updated Server Manager Console or PowerShell. When you configure NIC teaming, you select the physical network interfaces that you want to combine into a team. By default, the assumption if you are using VLANs is that you want to trunk them. However, you have the ability to specify a separate VLAN ID to assign to the team.

Guest VM Failover Clustering

In Windows Server 2008, Hyper-V provides the ability to cluster guest virtual machines with the requirement of iSCSI shared storage. While this is a proper solution, it presents some problems for customers that require the ability to cluster guest VMs, but that do not allow iSCSI storage in their environment. While Windows Server 8 continues to provide support for guest clustering through iSCSI shared storage, it also introduces guest clustering using a virtual fibre channel adapter. Leveraging the virtual fibre channel adapter, a virtual machine can have direct access to fibre channel storage via fibre channel ports, allowing customer to integrate into existing SAN devices.

Hyper-V Replica

Windows Server 8 also introduces a new continuous availability feature which is named Hyper-V Replica. Hyper-V Replica allows asynchronous replication of virtual machines between Hyper-V servers using only an IP network connection. There are no dependencies on the type of storage used or the workload that is running on the virtual machine. Replication is configured at the virtual machine level, so there is no requirement to replicate all virtual machines running on a server. Hyper-V Replica also integrates with Failover Cluster Manager so that the movement of a virtual machine between cluster nodes is known and replication can proceed without interruption.

Hyper-V Replica performs replication by tracking write operations on the source virtual machine and replicating those changes to the virtual machine replica on the destination server. Hyper-V Replica can be used over a LAN or a WAN. Replication is accomplished over the HTTP or HTTPS protocols to provide ease of firewall configuration. When configuring Hyper-V Replica, you have the choice to use integrated authentication without encryption of the replicated data, or certificate-based authentication which encrypts the data before it is replicated. Enabling and managing the replica status of a VM is possible using Hyper-V Manager, Failover Cluster Manager, PowerShell, and WMI.

Continuously Available File Server

In Windows Server 8, Hyper-V can leverage file shares as storage locations for virtual machines using the new SMB2 protocol enhancements. To ensure that file shares do not become a single point of failure, you can configure transparent failover of file shares between clustered file server nodes.

With Windows Server 8, you can build up to four node clustered file servers in active/active configuration that provide simultaneous access to a single file share across all cluster nodes. This configuration provides Hyper-V with VM storage that supports multipathing and the resiliency of a SAN, but without the management overhead of LUN allocation, assignment, and zoning.

Conclusion

While high-availability features have always been available in Microsoft Windows Server versions, Windows Server 8 significantly cranks up the stakes with networking, storage, and workload migration features that make continuous availability a real option across a wide range of cloud deployments. In Part 8 of this series, you will learn in more detail about continuous availability deployment scenarios supported with Windows Server 8.

If you would like to read the other parts in this article series please go to:

About The Author

Leave a Comment

Your email address will not be published. Required fields are marked *

This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.

Scroll to Top