Hyper-V, vSphere, and a Real World Look at a Project
It’s going to be a long time before the battle between VMware and Microsoft is resolved, particularly as Microsoft keeps improving Hyper-V and as VMware continues to add supporting services around the vSphere ecosystem. When it comes down to Hyper-V vs. VMware, there are definitely strong opinions on both sides of the fence – Hyper-V isn’t mature enough; vSphere is too expensive, etc. The lists on both sides are endless.
More Similarities Than Differences
As Hyper-V has matured, Microsoft had packed it with a lot of features that bring it very close to feature parity with vSphere. Sure, there are still differences here and there, but feature for feature, there are a lot of similarities between the products now.
I will tell you now that I don’t believe in simply copying and pasting recent “maximums” charts into my blog posts and then deciding which product is superior. While some items on the maximums charts are useful, the vast majority of them are just ridiculous for any kind of mainstream comparison. Sure, it’s great that that vSphere supports 120 SATA devices in a single VM. But, really… who cares? How many people are really going to stick than many SATA devices in a VM? There may be some outlier use cases for many of the maximums, but for mainstream purposes, the maximums themselves mean very little.
Cost and Ecosystem as Drivers
Fact: vSphere is less expensive than Hyper-V.
Fact: Hyper-V is less expensive than vSphere.
Wait, what? In reality, both are right, depending on the circumstances. The calculators provided by VMware and Microsoft are pretty much useless when it comes to any kind of real world comparisons. That said, as the products achieve feature parity, customers are looking for other ways to differentiate in an effort to pick the right tool for their needs.
Besides cost, the ecosystem – both outside the two vendors and within their own R&D departments – used to be differentiator between vSphere and Hyper-V. Although that situation is changing, VMware still retains a much broader ecosystem that Microsoft when it comes to the hypervisor directly. Further, it could be argued that VSAN, for example, gives VMware a competitive advantage over Microsoft (if you discount Microsoft’s Storage Spaces offering). VMware has realized that they cannot compete with vSphere alone and is spending a lot of time and money building products that revolve around the hypervisor.
Administrative Overhead Favors VMware
I work with both Hyper-V and vSphere. When it comes to initial administrative overhead – that is, the time it takes to deploy the environment and the complexity involved – VMware wins the battle. That said, Microsoft is catching up here, too, with the latest Hyper-V being far easier to deploy than previous versions. But, VMware still wins. It’s exceedingly simple to deploy vSphere and vCenter these days. vCenter can be deployed using a downloaded OVF and it just works. Virtual Machine Manager, on the other hand, takes a bit more effort. That said, VMM is not absolutely required software in a Hyper-V environment, whereas vCenter is a must-have.
A Case Study
Fact: Hyper-V is far less expensive than vSphere.
Let’s take a look at a project I recently completed for a client. The client’s vertical is higher education, which Microsoft supports very well. In fact, when looking at direct costs for both Hyper-V and vSphere, VMware almost never wins. Many colleges and universities are extremely cost-conscious on direct costs and many tend to de-emphasize personnel costs in TCO and budget calculations, so vSphere rarely comes out at the top of the heap on costs. Of course, VMware enjoyed first-mover advantage for a very long time, so many colleges and universities continue to happily use the product. Further, there are colleges that do value the benefits that VMware brings, as well as the ancillary tools, so VMware remains incredibly popular even against Hyper-V’s lower cost.
And then there are the ones that have just toyed around with virtualization and then decide that they need to finally take the full plunge. That’s where my project comes in.
When I arrived, the client was already pretty heavily virtualized. In fact, they were running Hyper-V 2008 R2 for a good chunk of their workloads. In addition, they were running vSphere 4 and vSphere 5 for other workloads (they bought vSphere a while back but were not paying S&S and did not want to foot the bill for a full vSphere deployment). They were also suffering from VMM issues, issues related to their Microsoft Data Protection Manager-based backup and recovery in their almost 100% Windows environment, and had a desire to implement more robust disaster recovery by leveraging their two remote sites. Finally, they still had a large number of physical servers, some of which were approaching 8 years old.
Yes, three hypervisors (two different vCenters) and lots of physical servers. A nightmare to manage!
They asked me to assess the environment and make recommendations with the caveat that budget was the key driver. As I’ve stated before, I believe in using the right tool for the right job and am comfortable with both vSphere and Hyper-V. In this instance, I made the following recommendations:
- Don’t bother trying to salvage any of the existing environment, with the exception of a couple of servers. Everything was either too old or underpowered.
- Buy four new hosts with lots of RAM. After doing a workload analysis, I determined that four hosts with 256 GB of RAM each would be more than sufficient for their needs.
- Implement Hyper-V 2012 R2 and VMM 2012 R2. Cost was the driver and for higher education Hyper-V and System Center are incredibly affordable. Manage the new environment with a brand new VMM 2012 R2 instance.
- Maintain Data Protection Manager, but upgrade to the latest version (DPM 2012 R2). DPM is a part of System Center, so it’s included in the bundle. Further, it’s surprisingly good as a backup tool for Windows-heavy environments.
- Re-use two existing Hyper-V hosts at remote sites. There were two servers that were reasonably new, so I recommended that they deploy one at each of their remote sites, with both running Hyper-V 2012 R2.
- Use Hyper-V Replica for DR. At one of the remote sites, I recommended that the client implement Hyper-V Replica as a way to replicate mission-critical virtual machines from the primary data center and use that data center as a DR hot site.
- Use DPM’s “chaining” capability for off-site backup. I recommended that the client use DPM at the primary site for ongoing recovery needs (i.e. a user loses a file) but then use what is known as “chaining” in DPM to copy those backups to remote sites. DPM operates by replicating only changed data (delta) as a part of this operation, so it’s not all that bandwidth intensive.
All told, we only spent around $80,000 on all of the new hardware and software needed, which included servers and storage (for backups). I estimated that the client would save around $200,000 over the next couple of years in hardware and software costs since they wouldn’t have to replace their existing physical servers as they failed. They had staff that was already comfortable with Hyper-V, VMM, and DPM, so retraining wasn’t necessary. This project vastly simplified the operating environment, making things far more efficient than they once were.
The decision to use Hyper-V was far easier to make than it would have been a couple of years ago. Hyper-V has reached a maturity level enabling its use in far larger projects than it would have been in the past. In certain verticals, Hyper-V enjoys a massive cost advantage. That said, vSphere remains the tool of choice for many and enjoys benefits related to administrative overhead that Hyper-V simply can’t match yet.