Interview: IDC Research on the Growth of Virtualization

By Charlie Leffingwell
published: Monday, October 15 2007

The following is the transcript of a podcast interview with IDC’s Stephen Elliot, originally posted on August 17, 2007. To listen to the original recording, click here.

SE:Well, it’s certainly a very hot topic and these are questions we often get from enterprise IT organizations. There are a couple key benefits.

First and foremost the obvious one is cost savings, specifically with server infrastructure and the reduction of server hardware.

Second benefit is really around the process side. Enterprise IT organizations often see process improvements specific to availability problem management, change management. These are areas of control that in the distributed world have often caused angst for many of us. So what we’re seeing now is a streamlining of the definition of what these changes and integration of problem management into virtual infrastructure.

A third key area – from a benefit standpoint – is efficiency and becoming more agile to business requirements. One example we often hear about, and in fact advise users on, is provisioning. With virtual machines and the ability to provision those machines, we’re seeing tremendous time savings in relation to the need to meet business objectives and business requirements for innovative new IT services.

Another benefit we often hear about is around this notion of an architecture decision. More and more we’re hearing some of these large enterprise IT organizations looking at virtualization not just as a tactical cost savings opportunity but something that is an architecture decision for the technology. From an organizational structure, we’re hearing CIOs start to create virtual computing teams to help drive the proliferation of the technology across not just server but storage, application and desktop organizations. The key benefit for that is really executive level support and buy-in for looking at benefits across multiple silos and multiple teams to drive out strategic advantage. This is often through more agile infrastructure.

As for the challenges…there are certainly some key ones. Now, first and foremost there’s still a learning curve, albeit a lot less than it used to be just two or three years ago. People have to get comfortable with what’s the technology, why should I care and certainly whether the strategic opportunities not just across technology infrastructure but across the business requirements. Another challenge is understanding the application of the technology across different silos. And I think this is not just a learning curve but challenges across integrating existing processes that may be attributed from the IT infrastructure library or COBIT or even Six Sigma. And I think looking at these processes, whether its problem management or change management, identifying these key challenges and then making sure they’re a part of the virtual infrastructure. It’s an opportunity because it can certainly streamline some of these processes but it’s also a challenge because sometimes these processes aren’t fast enough to respond to IT’s requirement to deliver faster virtual machines or more services based off of these machines.

Another key challenge is performance. Let’s face it – this is technology that’s not, right now, right for every application. There are IO and performance limitations on it currently. They’re being addressed by the vendor community. Certainly long term this might change but currently file, print, Web services are often X86 platforms virtualized. So the challenge is, from the performance side, to really up the ante to be able to offer more mission-critical applications.

Finally, I think a key challenge is just really getting executive-level buy-in and visibility. This has noticeably increased over the past year but for most cases is still viewed as somewhat tactical and I think that really elevating its awareness, the benefits and certainly the opportunities for broader IT organization is an ongoing challenge.

SE: One of the things we’ve seen increase over the past six quarters is this notion of “How do I manage virtual infrastructure?” Obviously this is a big challenge, as we’ve seen over the past 10 or 12 years IT organizations really put a lot of resources, staff training, and a lot of emphasis around getting control over the distributed environment. Not just in terms of change management, problem incident release management but certainly in relation to moving from a component perspective on infrastructure to really a service perspective. That change requires organizational change; it requires a cultural change within IT, really moving from a component to a service culture.

For virtual infrastructure we’re seeing this as a key challenge because obviously it’s pretty easy to spin out virtual machines and create VM sprawl as we call it. But to really challenge, to really manage the problem and changes and configurations to the point that IT organizations are comfortable with putting more and more virtual machines into production and more applications onto those infrastructures. Everything from monitoring change management, physical to virtual conversion, virtual machine access control, performance and resource management…these are all critical things that users have to consider when they’re building out virtual infrastructure.

Another key consideration is virtual machine migration, being able to automatically restart these machines and really drive some level of automation in terms of task execution from scripts. There’s really a plethora of challenges that don’t really go away…and in some cases get easier and other cases get a little harder. But the net of it is that IT organizations really need to focus on this discipline and make sure that as they continue to spin out virtual machines they’re able to really control the environment, manage it effectively to the IT services that are dependent on the virtual machines.

SE: I think there are a couple key recommendations that we make to IT organizations. First and foremost, when you think about virtualization it’s not a panacea for all your management challenges – it’s not right for every environment. So first and foremost really identify who within your organization is driving the budget, driving the decision. And secondly, really think about not just the tactical side of the discussion but certainly the strategic side, and assess if you’ll be able to garner benefits not just across the server side but the desktop; the client side, potentially the storage side with business continuity processes, and to really determine if it is a tactical decision or is it something that may be a strategic architecture decision.

A second piece of advice is really around the management challenges. And this is around not just the process understanding but certainly the product implementation as well as the need to understand processes operational as well as application development processes in relation to the virtual infrastructure.

A third and important piece is managing expectations because there’s no doubt that some areas when you’re looking at virtual machines, areas such as provisioning, offer an opportunity to show how fast you can off IT services and really drive out efficiencies. We’re advising users to really think about how they can import service level agreements as well as metrics – it gives them some cushion because as they get more time back and certainly can offer faster provisioning don’t be so aggressive that you put yourself back in a hole. There are wonderful opportunities to become a profit center through smart chargeback models as well as really make yourself a hero in terms of faster provisioning cycles. But give yourself a cushion when creating service-level metrics and make sure that you set the proper expectations so that you can be successful longer term.

SE:At the high level there are a couple key things. First and foremost, really listen to the install base – not just the midmarket folks but certainly large enterprises in critical fashion and how they want to build out these virtual platforms and the broader opportunities and challenges associated with it. Take that into account and continue to innovate the product lines.

A second thing we advise the folks on at a high level is really looking at it as an opportunity to drive out not only new revenue streams but to really think about how the technology can be applied and help drive out faster implementations or lower costs of operation across IT organizations. So whether it’s just a server discussion or storage discussion or a client desktop discussion, really try to bring these pieces together not just in terms of the core hypervisor level, but particularly around the management challenges. Think in concert so IT organizations can get more comfortable with the adoption as an architecture decision and really help solve problems specific to performance, availability, virtual machine migration and movement, and access control and even physical to virtual conversion. Lots of different pieces of management that need, you know, desperate attention aren’t necessarily getting it today in a strategic fashion.

Finally, let’s face it – there’s not a single vendor that can solve every problem here, so create tight partnerships, acquire when necessary, but really have a lifecycle approach to this. Really think about how virtual machines go through the lifecycle in production, the key consulting services around chargeback and capacity planning that often can be required, and finally, really think strategically about what this means in meeting business objectives. Tie this back to the line of business and business processes that are ultimately driving our decisions. I think a lot of these areas are big challenges for vendors. They’re a work in progress, but generally speaking some interesting small steps forward to help organizations make the most of the technology.

SE: We’ve done some great research around this and have talked to many large production environments that are using virtualization in various capacities often looking at large production, virtual machines, and assessing challenges and this notion of benchmarks. One of the things we decided to do was talk in depth about what they’re experiencing and what are the different metrics. We often see a couple key things driving decisions and helping folks understand what they should expect as they build out their production environments with virtual machines.

One of the key things we found is this notion of staffing. Surprisingly, we found that IT leaders should not really expect a drop in the number of administrators when they utilize virtual infrastructure. In fact, there’s an opportunity to reallocate that time that they get back and/or reduce the need for additional hires. It’s not so much a cost cutting mode per se – as it relates to staffing. It’s more about getting time back in your pocket that you can reutilize for strategic projects and it also helps reduce the need for additional hires. So essentially your server infrastructure goes up, your demands go up, your hires stay flat.

Another key benchmark is the virtual machine to administrator ratio. Often IT organizations talk about server to admin ratios. This really turns that on its side – what’s the VM tab in ratio? What we’ve found is the average number of virtual images per admin is about 200:1. That’s a reasonably sophisticated chop, but it’s not out of the range100 to 200 is something we’re finding relatively consistent within enterprise accounts.

Another key thing is looking at the organizational impact. We’ve found that about 15% of IT organizations are creating a specific virtualization group, sometimes called the virtual computing team or something like that. These are folks that often have made a pretty strategic architecture decision. Now, the 85% of IT organizations are often allocating ownership of virtual infrastructure to a savvy group within the server and/or storage team. So they’re creating virtualization specialists or just adding this task on to an existing set of high performing staffers.

And finally, on the process side we found that ITIL (IT Infrastructure Library) within IT operations is clearly gaining in adoption for the global 2000 and we’re also seeing that ITIL is often part of virtual infrastructure and the management workflows that are often required. These are problem, change, incident management as it relates to the virtual machine. Sometimes ITILs not fast enough but, nonetheless, it gives people a good foundation for establishing these management workflows within virtual infrastructure. Those are some critical benchmarks we’ve found.

SE: : I think from a management perspective, first and foremost it’s about making sure your compliance, your change management, and your policies are intact, organized, created, defined in terms of what the business requirements are and I think that’s going to be an ongoing challenge as you start to get folks who look at mission critical applications onto this platform and really drive out business alignment through this faster, provisioned, more agile infrastructure.

Now, from a product standpoint, I think the marketplace is going to show some rather innovative packaging, shipping, some innovative pricing, that’s really going to drive out the adoption of different types of tools that support the core hypervisor layer today, and certainly in terms of management tools that come out…these are going to develop more over the next 2 or 3 years.

I think from a process standpoint, it’s critical that companies really consider what the core processes are that they have on the physical side and consider the adoption onto virtual infrastructure.

It’s not going to matter is it physical or virtual. What’s going to matter is how you really relate to, communicate to the business impact of the services that you’re delivering and how do those really integrate into existing process workflows. A singular dashboard for physical and virtual environments is probably, within three years, going to be the norm. When you take the management workflows, the products, and the processes together it becomes a rather dynamic platform that IT, that we, can really take advantage of to drive out more business agility and, from a CIO perspective, really have a solid impact on retaining customers, driving higher margins, driving new innovative business services…certainly making IT more of a profit center that’s service based versus a cost center that’s more component oriented.

I think if you take these into account you’re looking at a very interesting recipe for many IT organizations moving forward.