The most significant disruptions come when the thing you’ve always just taken for granted changes. In November 2023, Broadcom completed its acquisition of VMware. At $69 billion, this is objectively causing a massive upheaval on a global financial scale. In IT terms, this ranks just slightly behind Dell’s acquisition of EMC back in 2015. This change was big to be sure — but disruptive? Servers, PCs, and storage is a vendor space so commoditized that even back then, rotating between them was basically standard just to keep prices low.
How Does the Broadcom Acquisition of VMWare impact users?
Today, we’re talking about our hypervisor. Let’s be honest with ourselves; when is the last time you sat down with your technical team to seriously (re)consider your choice of hypervisor? Some more eager souls may have tried to start contingency planning immediately following the announcement in May 2022, but details at that time (and in the following 18 months) were exceptionally scarce. During this time, speculation and prognostication from analysts was rampant (sorry), but there weren’t a lot of hard facts. Now, the picture is finally taking shape, and its becoming clearer and clearer. This acquisition is the biggest and riskiest disruption the I&O space has seen over the last 15 years.
I can practically hear the chorus of readers disagreeing with me. And yes, this disruption is even riskier than the cloud. Sure, some early adopters ended up spending untold sums for mediocre results and there were data leaks since S3 buckets were put on the internet without passwords, but the fact remains that going to the cloud was still a choice. FOMO was the biggest driver of people adopting this change 15 years ago. Infrastructure & Operations (I&O) leaders explored, tested, and stumbled through new ways of working, either at their leisure (or at the behest of their CIOs). As a result, organizations have taken a dozen years to move a third of their workloads to the public cloud. Today the driver is far less abstract, and VMware by Broadcom prices are generally increasing. Compound this with the countdown to the expiry date on your latest license agreement — a year or maybe two for those who were lucky enough to lock in back in May 2022. So, I ask you, chorus of readers — what do you think is riskier?
- Investing some money and time to test workloads in a new cloud environment?
- Finding potentially substantial increases to your historic budget just to keep the lights on?
- Planning and executing the migration of your entire on-premises infrastructure stack within a year or 18 months?
They say being a leader means making the hard choices. Today’s I&O leaders get to choose between spending way more money this year than the last — potentially with no corresponding increase in value — and crowd out other more innovative priorities from the budget. That, or they might choose to lift and shift an infrastructure stack that’s been organically growing for 15+ years to something new that requires new skills from your staff. The only upside is that risks can be mitigated. Reputationally, I&O has a well-earned (if sometimes slightly over blown) aversion to risk. Risk mitigation is in our blood. So, let’s start brainstorming about what we can do.
Planning for a Hypervisor Migration
For the purpose of this exercise, we’re going to assume you’ve chosen not to just pay the increased licenses fees — not that there’s anything wrong with just writing a check. Nice things cost money and sometimes it’s easier to just pay. Of course, there are no guarantees that a vendor will continue to invest in product development or customer support. In fact, there are several historical models that suggest quite the opposite. Paying, however, does remove the urgency of having to get a large project done in 18 months.
There are many moving parts in an IT strategy: Application rationalization, cloud migration, workload replatforming, and refactoring. There is never enough time to do all the things the CIO wants, and buying time is a perfectly reasonable option, especially if that choice means I&O can pass along increased costs to those other teams. A little financial pressure can do wonders to encourage adherence to deadlines.
If you’re not going to pay, though, then you’re going to have to move. So, the questions here are where and how you’re going to do it.
- Public Cloud IaaS
Lift and Shift had its heyday 10 years ago before CFOs started looking at their monthly bills, but thanks to recent events, this may finally be a cost-effective option. Offerings like Azure Stack and AWS Outposts even let you straddle between both worlds.
- Hyper-converged Players
Nutanix may not have lived up to its early hype, but it has survived going public. Proxmox has been around for about 15 years and has seen twice the traffic to their website in the last six months. Scale (who is offering to buy out your remaining VMware term if you switch) and StarWind round out the original surviving HCI players.
- Container Focused
If containers are your future, there are pioneers like KubeVirt that can handle both pods and virtual machines (VMs). And, of course, everyone knows OpenShift. RedHat is certainly easier to deploy today than in previous years, but if you’re still coming up short in the skills department, you can work with companies like Canonical for 24/7 management.
- Vendor-managed Solutions
Does your operations team no longer want to bother with caring for and feeding racks upon racks of infrastructure kit? HPE GreenLake and Dell APEX both offer as-a-service solutions to fit some of these needs and all your budget.
- Cost-effective and likely in your datacenter
Microsoft Hyper-V. This was never as fully featured as VMware, but what some have learned is that they never needed the best-in-breed after all. Sometimes good enough is actually better, and chances are, most of you already own this.
Post Migration Considerations
Insult adds to injury when you realize that, with hypervisor migration, it doesn’t stop once you’ve picked the place you want to go. How will you monitor that place? How do you back it up? How do you instantiate new workloads or change the workloads you already have? All these problems you’d thought you solved and were finally done with are now up for debate again. Well, good news is, this is one area that we may finally catch a break.
Regardless of where company data resides, it’s up to I&O to protect it. 25 years ago, this meant primary and secondary data center sites. Then users started keeping valuable information on PCs, and then came the rise of local network attached storage (NAS) in field offices. Then came the cloud, which was just like another data center location. Today the cloud has exploded, with the average enterprise consuming dozens of platforms and hundreds of SaaS apps. Still, no matter where the company data exists, it’s up to I&O to protect it. So, they turn to vendors. Modern data protection software has been engineered to back up enterprise workloads in the data center, on the local PC, on the NAS, in the public cloud, and in even an SaaS app. Since vendors want to write software once and sell it to many, they’re already incentivized to support multiple SDDCs, hypervisors, HCIs, vendors, and public clouds. In short, the data protection software you’re already using likely has the ability to protect data in the new hypervisor environment you’re moving to!
Once you have figured out where you’re going, how you’re going to get there is the next problem. Again, the risk-adverse I&O leader may have caught a break. Cloud backup was an early and persistent win for public cloud IaaS. It made it easy to follow best practices, storage prices weren’t outrageous, connectivity was supplied via cheap DIA circuits, and you didn’t have to worry about offsite storage facilities or aging and incompatible LTO drives. The championship trophy, however, was awarded when the public cloud could effectively be used as a disaster recovery (DR) failover site and as a restore target. Sure, IaaS VMs are expensive, but not as expensive as all the redundant kit that’s required to outfit a secondary cold site. The same tried, true, and trusted data protection software also had the ability to take workloads from the data center and restore them to the public cloud. This type of portability is what we need today.
Planning and successfully executing a hypervisor migration is not an easy task, nor is it one I’d wager that many I&O people (unless they were early Nutanix fans) have had to do during their careers. GSX launched in 2001 and VMUG started up in 2010. VMware was a market leader in product innovation, consistently got high marks for customer support, and, despite some price increases along the way, their trajectory was ever upwards. No one really practiced or planned for this though, and now you have limited time and many unknowns to work through.
The good news is that sound data protection practices — and its associated technologies — have mitigated some of the risks that come with migration. Going to a new hypervisor will mean exercising muscles we haven’t used for perhaps 20 years with experimenting, trial and error, breaking in new vendors, running through use cases and processes and R(ing)TFM. The heady historic days of in-place upgrades were only survivable because we had confidence in our backups. Data protection was and still remains the best mitigation we have against a resume-generating event.
For more practical tips on hypervisor migration, watch our on-demand webinar Hypervisor Migration: Keeping Data Safe in a Changing Industry.
