Watch out, VMware. Redmond wants a piece of the cloud market, but hypervisor improvements in Server 2012 can also boost reliability and DR in smaller shops.
In enterprise data centers, VMware vSphere is to hypervisors what Microsoft Windows Server is to operating systems: the undisputed kingpin. The numbers don't lie: Our latest InformationWeek ReportsVirtualization Management Survey found fully 90% of respondents using some version of vSphere as a primary hypervisor platform, with Microsoft's Hyper-V a distant second. In our Windows 8 Survey fielded in June, just 14% called Hyper-V their main production hypervisor versus 23% saying they will continue to scale out vSphere/Xen/VirtualBox because they don't trust Hyper-V.
Microsoft plans to change that perception.
At last week's Hosting and Cloud Transformation Summit, Scott Ottaway, Microsoft's worldwide industry director for service providers, made clear that the company is no longer content to play second fiddle in the realm of virtualized infrastructure, and it's going to battle with a significantly revamped Hyper-V. While Ottaway was speaking to a crowd of managed service providers, colocation providers, and data center technology merchants, it's clear that the hypervisor technology in Windows Server 2012 is equally beneficial to public and private clouds alike, so expect a similar Microsoft marketing push to enterprise IT.
Ottaway's take on Server 2012's top new features illustrates how Microsoft has been busy plugging holes in a previously serviceable product, but one that invariably lagged VMware's technology: improved virtual machine scalability and performance; network virtualization with quality of service; NIC teaming; disaster recovery with Hyper-V Replica; shared-nothing live migration, such as between nonclustered systems with only local storage; VM failover prioritization; and live storage migration.
It's a heady list. One look at Hyper-V's new capacity specs [PDF] shows that when Ottaway says Microsoft has the goods to allow hosting providers to deliver cloud infrastructure at lower cost (presumably compared with VMware, although he kept the ad hominem comparisons to a minimum), he's not just blowing smoke. Over a range of parameters, from the maximum number of virtual CPUs per VM to memory limits per VM to the maximum size of virtual disks, Hyper-V matches, or in most cases, exceeds its main competitor.
Now, that's all well and good for infrastructure-as-a-service providers looking for a less-expensive alternative to vCloud, but how many enterprises are putting 4 TB in a server in hopes of running 320 VMs? For your typical IT department, particularly small shops, the real enticement with Hyper-V 2012 isn't massive scale. It's improved reliability and redundancy.
Server 2012 now fully decouples VM management from the physical infrastructure and allows nondisruptive migration, simultaneous or queued, of multiple VMs from one host to one or more alternates. For example, you could free up resources on an overloaded server hosting six applications by targeting one application to another smaller machine and two others to a third midsize system. And such sophisticated live migration features don't require a SAN--VMs can live on local file shares. Furthermore, Server 2012 can move virtual hard disks attached to an active VM without disturbing the running applications. Together, these features give IT more flexibility in placing and relocating VMs with no downtime. Whereas the new and improved migration features can improve reliability and resource utilization within a primary data center, Hyper-V Replica [PDF] provides an easy, automated way to replicate VMs to a secondary site.
Let's state up front that HVR isn't designed for large organizations that already have an expensive SAN-based replication system. It's squarely aimed at small and midsize companies looking to get some semblance of a DR system by, for example, having a secondary computer room at a branch office host synchronized copies of critical virtualized applications. All you need is HVR running at each location and a WAN connection--that's it. In fact, you don't even need a full-blown copy of Windows Server 2012; Microsoft has a standalone bare-metal hypervisor, Hyper-V Server 2012, as a freely available download.
In another boon to smaller shops, the servers needn't be identical and can use any type of supported storage; HVR takes care of the rest. As Aidan Finn describes on his Hyper-V blog, HVR works by maintaining and asynchronously replicating log files from one host to another. These logs are then replayed on the target system every five minutes. Since replication is asynchronous, you don't need a high-bandwidth, low-latency connection between the two sites (although busy servers and slow networks may mean you'll fall behind the five-minute log replay window). Fortunately, since you're just replicating change logs, it's unlikely the data rate will overwhelm even broadband branch office circuits. Another nice touch is that HVR can be configured to store multiple recovery points, any of which can be selected when failing over to the secondary host.
One big caveat: Failover and recovery is a manual process. When the primary site goes down, IT must kick into action. Although failover can be managed with Hyper-V Manager, a better option is to automate the process using PowerShell. In fact, to bootstrap creation of automation scripts, Microsoft has developed a library of Hyper-V cmdlets for controlling virtually every aspect of the hypervisor, including most replication features.
Circling back to Ottaway's target audience, MSPs, gives a clue of a Hyper-V deployment option that could be even more interesting to IT: a hybrid private/public environment. Suppose you don't have a branch office or one with suitable space for spare server capacity. Lease a Windows Server at an MSP and fire up HVR, then rest assured that not only are your critical apps automatically protected, but they are being replicated to an environment that's probably more reliable and better monitored than your primary site.
Given the massive overhaul to Microsoft's virtualization platform--we haven't even touched on features of greater interest to large enterprises, such as NIC teaming, network QoS, VM failover priority, and concurrent live migration--don't be surprised if our next virtualization survey shows some sizable erosion in VMware's share. Who knows, if Microsoft's pitch is successful, Hyper-V just might be the IaaS foundation for your next cloud deployment.
Multicloud Infrastructure & Application ManagementEnterprise cloud adoption has evolved to the point where hybrid public/private cloud designs and use of multiple providers is common. Who among us has mastered provisioning resources in different clouds; allocating the right resources to each application; assigning applications to the "best" cloud provider based on performance or reliability requirements.
InformationWeek Must Reads Oct. 21, 2014InformationWeek's new Must Reads is a compendium of our best recent coverage of digital strategy. Learn why you should learn to embrace DevOps, how to avoid roadblocks for digital projects, what the five steps to API management are, and more.