Why Virtualization is a Requirement for Private Cloud Computing

Like candy bars, it’s just a lot less messier than the alternative.

Caramel. Chocolate nougat. Coconut. No matter what liquid, flowing, tasty goodness is hidden inside a chocolate bar, without the chocolate shell to hold it we’d be in whole a lot of trouble because your mom would so be on you for that mess, let me tell you.

Every food-stuff that is liquid or gooey or both is encased in some sort of shell; even the tasty Swiss cheese and prosciutto hidden inside chicken cordon bleu is wrapped in, well, chicken. In many cases the shell isn’t even all that appealing, it’s all about getting to the center of that Tootsie Roll Pop as quickly as possible.

And that’s really what enterprise IT wants to do, too. They want to get to the heart of things as quickly as possible and as efficiently (and cost effectively) as possible. That’s why, in a chocolate-shell, (sure, nutshell would be more appropriate but it doesn’t fit well with the analogy, does it?) virtualization might as well be a requirement for the implementation of a private (on-premise) cloud computing initiative.

RESOURCES are FLUID

If you look at what you’re managing with cloud computing, it’s all about resources. RAM. CPU. Storage. Network. Those are the components that are fluid and without some sort of shell in which to “wrap them up neatly” they’re just going to make a big mess all over the data center. Now, cloud computing providers – many of whom were in business long before virtualization became “the one true way” to manage those fluid resources – may or may not leverage virtualization a la VMware, Microsoft, or Citrix to enable the management of the resources which make up the foundation of their dynamic, cloud computing environment. And they almost certainly don’t leverage any of the modern orchestration systems to manage those virtual machines (if they use virtualization) because, well, they didn’t exist when most providers built their implementations.

And it’s unlikely that providers will use those frameworks in the future because they tend to have more stringent needs in terms of multi-tenant support and integration with metering and billing systems. Additionally, they need to provide a method of management via an API and that’s almost always customized for their specific environment.

An enterprise, however, is very different.

VIRTUALIZATION = CHOCOLATE SHELL

While it’s certainly the case that virtualization is not a requirement for cloud computing, for the enterprise it’s as close as it gets. Without virtualization the systems and frameworks required to package, provision, and easily manage applications and their associated resources would require so much investment and time that ultimately it would take decades for them to implement and subsequently realize the benefits.

Virtualization therefore provides that all important encasing, chocolate shell in which to deploy applications and resources and to integrate those packages into a broader automation and orchestration framework. That framework may be custom-built or it may be based on the management offerings coming from the virtualization provider (a la VMware vSphere and vCloud and Microsoft Virtual Machine Manager, etc…). The enterprise needs the “package” because in most cases it simply doesn’t have the time to build it out themselves. Worse, the support and long-term management of such custom systems impinges far more on the realization of benefits than even developing the initial packaging system.

Leveraging “industry standard” - and nearly commoditized at this point - virtualization offerings also affords enterprises mobility and a growing selection of third-party management and migration tools that leverage the APIs and frameworks of those virtualization offerings. The chocolate shell is fairly standard, even if its contents are customized for the enterprise. It is familiar, recognizable, and there are an increasing number of folks in the field with expertise in managing such products.

Enterprises have other things to do with their time, like integrate applications, analyze data, troubleshoot the network, and develop new applications and features for existing applications. Providers don’t have these distractions, they have the time to build out whatever they’d like under the covers and abstract that into a nice, neat API that’s easy to consume. That’s their line of business, their focus. But the enterprise? Cloud computing and automation is a means to an end; a more efficient, automated data center, a la IT as a Service. Virtualization, especially industry “standard” virtualization solutions that are well supported by an existing and dynamically evolving management and tool ecosystem, gives enterprise IT a wrapper-ready solution that’s easy to consume.

AN ENABLING TECHNOLOGY for the ENTERPRISE

Perhaps “requirement” is too strong a word, as there will always be those enterprises large enough and diverse enough and with large enough budgets to “roll their own”. But virtualization is – or should be – a best practice for any enterprise considering the implementation of a cloud computing environment. Without it, organizations will be left to their own devices (pun intended) for most related solutions – such as management and integration with charge-back/metering services – and consequently will either take significantly more time to reach parity in terms of return on their investment. If they ever get there.

It is also a requirement in that virtualization of applications results in the ability to more easily and affordably entertain the possibilities of new architectures, such as those proposed by Dave McCrory in “The Real Path to Clouds.” His “phase 3” architecture is a ways off in terms of implementation, but just moving to phase 2 (scale out at both the web and application server tiers) can be problematic for larger organizations. Virtualization affords the ability to “play with” (for lack of a better way to say it) new architectures without significant investment and risk to the availability and business continuity required of production systems. Consider that it is far easier to share resources across both the web and application server tiers when in a virtual form factor, thus making it possible to adjust, dynamically, the resources necessary to sustain availability and performance requirements during spikes or peak usage periods.

So while virtualization is not a requirement for “cloud computing”, in general, it is for most enterprises who are planning on moving forward with cloud computing and data center automation and may be interested in new architectures that are better able to adjust to hyperscale requirements.

Related blogs & articles:

Published Nov 08, 2010

Version 1.0