A few years ago, I wrote a controversial column listing nine reasons Hyper-V was not enterprise ready and suggesting that Microsoft had lost its innovative edge. I think the only Microsoft employee who didn't send hate mail was Bill Gates, and I still maintain that column cost me an MVP award.
While I stand by my previous assessment, I also maintain that the situation has changed significantly. Over the past two years, Microsoft has gone strongly into innovation mode, not only on Hyper-V but on many other product lines. The turnaround started with Exchange 2010 and its leveraging of local storage as opposed to complete reliance on expensive shared storage and extends from System Center 2012 (probably my favorite) to the new version of SQL Server, which is following in Exchange 2010's footsteps and technology, to the new, long-awaited and overdue version of App-V, a technology Microsoft acquired many moons ago but on which it bestowed very little development effort.
But back to Hyper-V. The new version not only addresses all of my previous beefs with the product, it goes from a position of just playing catch-up with the market leader to actually giving VMware a run for its money at the feature level. This is the first release where Microsoft is putting forth a feature that VMware does not have: the ability to do live migrations of storage virtual machines.
Let's take a look at that and eight more features that earn Hyper-V a serious look.
1. Storage Live Migration: This capability is now built into Hyper-V Manager as opposed to requiring System Center Virtual Machine Manager, as was the case with the Quick Storage Migration. Storage live migration allows IT to migrate a VM, without any downtime, from one storage system to another. Remember that innovation I was talking about? Well, traditionally with storage live migration technologies from Microsoft, VMware and others, a shared storage repository is required for this feature to work properly. In Hyper-V 3, that is not the case. While you could use shared storage, of course--and I highly recommend doing so--you can migrate a live storage VM from local disk to local disk without any downtime. Now the ball is in VMware's court to match that functionality.
2. Concurrent Live Migrations: I have for many years criticized Hyper-V's lack of concurrent live migrations, and I'm very happy to report that the new version finally supports this capability. For a virtualization administrator, this is invaluable functionality. We live in fast times, and we need to be able to react at the speed of the business. Quickly moving all the VMs running on a given server is a definite requirement in any virtual infrastructure, and this release delivers that.
3. Dynamic Memory: While not a new feature in Hyper-V 3 (it was available as of Hyper-V R2 SP1), it's worth noting in a list of reasons Hyper-V 3 is ready for enterprise use. In a nutshell, dynamic memory is a memory management enhancement that allows IT to automate adding or removing memory from a VM on the fly--very helpful when trying to improve the density of VMs on a host, for example. And it's a vital feature in any enterprise virtual infrastructure.
4. Continuous Availability: This is actually a collection of technologies in Hyper-V 3 that includes, in addition to Live Migration and Storage Live Migration, NIC Teaming and Guest Failover Clustering.
-- Failover Clustering: Today, the cluster supports only 16 nodes; in Hyper-V 3, the cluster will be able to support 64 nodes and as many as 4,000 VMs.
-- NIC Teaming: IT can now combine NICs from different vendors, say, Intel and Broadcom. We also have three modes for configuring NIC Teaming: switch independent, static teaming and Link Aggregation Control Protocol; LACP is huge as it extends support for demanding applications like Citrix Provisioning Services.
Finally, for Windows Server 8 or Server 2012, depending on what the name ends up being, Hyper-V 3 has a really cool feature that leverages SMB 2.2 (I am super-excited about SMB 2.2). It can leverage file shares as storage destinations. I'm sure you're thinking "single point of failure," but remember, you can build up to four-node active-active clustered file servers, which provide simultaneous access to file shares. Yeah, SMB 2.2 is cool; the locking mechanism is great as well--watch out, NFS.
5. Network Virtualization: Microsoft is all-in on cloud, and in order to be effective in the cloud era, you need the network stack in your virtual infrastructure to be solid. It's worth mentioning here that Cisco supports Hyper-V on the Nexus 1000V, so the ecosystem is also coming together. In addition, Hyper-V 3 will support policy-based, software-controlled network virtualization; this is crucial in the cloud era because everything will be about policy-driven automation and orchestration, all key enablers of infrastructure-as-a-service deployments. As part of the Hyper-V 3 network virtualization capabilities, you can also create a bridge between your on-premises and cloud deployments that enables you to move your subnets into the cloud and create logic to allow them to communicate, essentially creating a hybrid cloud.
6. Storage Enhancements: No enterprise virtual infrastructure is complete without tight integration with storage, and Hyper-V 3 introduces some impressive improvements here as well. First, the new Offloaded Data Transfer is similar in functionality to VMware vSphere APIs for Array Integration, and I'm very eager to see how that improves or even solves the locking issues with CSV, which still redirects I/O through the parent partition. Virtual machines can now support up to four vHBAs with direct access to SAN LUNs using multipath I/O. You also have built-in replication, hardware snapshotting and, my personal favorite, Remote Direct Memory Access networks for SMB storage
7. Platform Enhancements: The platform has seen some major improvements as well, with support for 320 logical processors and up to 4 TB of memory per host. It's now possible to provision virtual machines with up to 64 vCPUs and 1 TB of memory, a huge upgrade from four vCPUs and 8 GB of memory. The new VHDX file format supports up to a 16-TB virtual hard drive. These enhancements will fuel the virtualization of Tier 1 applications and are critical for an enterprise-class virtualization platform.
8. RemoteFX: This, again, is not a new feature of Hyper-V 3, but it's very relevant to enterprise IT. Hyper-V supports GPU virtualization, which in desktop virtualization applications can be of great benefit in terms of enhancing the user experience. Essentially, you're able to expose a virtual graphics device to a virtual machine and allow multiple virtual desktops to share a single GPU. This would enable users to run graphically intensive applications on a VM.
9. Hyper-V Replica: Hyper-V Replica is a new feature of Hyper-V 3 and is somewhat comparable to VMware vSphere Fault Tolerance. Hyper-V Replica will asynchronously replicate virtual machines from one Hyper-V host to another over an IP network. The process is configured at the VM level, so it's not an all-or-nothing proposition. The technology tracks write operations on the source machine and replicates them to the destination VM so that both VMs are in constant lockstep. If one VM fails, the replica takes its place without missing a ping--a pretty cool enterprise-class feature.
I am really excited about Hyper-V 3, and I hope Microsoft continues its innovation trend and directs more attention toward alternatives for the parent partition approach and to building a better clustered file system. CSV is not bad, but I think the natural evolution is a really solid, scalable file system. Hyper-V 3 will be the first real challenger to VMware vSphere 5, so let's see how VMware responds. I think competition in this space will continue to drive innovation, and the customer will definitely be the ultimate winner.
Elias Khnaser is the CTO of Sigma Solutions, a vendor-agnostic systems integrator focusing on mobility, cloud and big data. Follow Elias on twitter @ekhnaser.