I’ve spent a lot of time lately in front of customers trying to help them understand what cloud technology means and, more important, how it can help their businesses be more efficient, effective and better able to meet or exceed their goals and objectives. In most cases, prior to speaking, the reactions I see are vacant stares and furrowed brows (sometimes on the same person…which is just a little disconcerting) as there is just way too much conflicting information that these businesses have to deal with. In today’s technology landscape, at least the one where some tech company is trying to sell you something, the overriding buzzword is cloud and it’s attached to just about everything and anything.
Traditional Datacenter Versus ‘68 Dodge Dart
In order to break down the many different concepts of cloud and cloud technologies, let’s compare a traditionally managed datacenter with a 1968 Dodge Dart. Now the Dart was a workhorse of a car back then, especially the 1968 version when it was available with a normally aspirated (it had a carburetor) 318 cu in 2-barrel engine rated at 230 bhp. Nothing special, nothing fancy but that’s what made it the go-to car for a newlywed’s first car, graduation presents and a family’s second car…simply, it was bulletproof. The Dart could be taken to any mechanic and he would know exactly what was under the hood and likely had a part that fit. Note that this is very much like a traditional datacenter today…nothing special, nothing fancy but pretty much bulletproof and any admin or engineer walking down the racks & stacks knew what was happening and how to fix whatever fell down.
Virtualization and Fuel Injection
Now let’s go forward in time to the last 5-6 years as virtualization starts to take hold and transform datacenters from how many workloads run per machine to how those machines are managed. If we make a comparison of the transformation of a traditional datacenter to a virtualized datacenter (my definition is at least 40%+ virtualized) to the ’68 Dart, it’s the same as if we took that normally aspirated gasoline engine (leaded of course) and converted it to turbo-charged and fuel injected with a catalytic converter. What we’ve done is upgraded existing components to be more efficient and effective versions (and some supporting components) but not much else is changed. Suffice it to say that the infrastructure (engine) is now as efficient and effective as possible but that the same guys who worked on it before are still able to work on it now. OK? Does that make sense? In a virtualized infrastructure, the same admins and engineers who were responsible for maintaining the systems before are the same guys who do it now (with some additional training and basic tool changes) just like the same mechanic who worked on your Dart before can work on it now.
Gas Turbines and the Cloud
This is where it gets interesting….suppose you now want to take the Dart to another, altogether different level. One where you wring out hundredfold gains in performance and the absolute most efficiency from the fuel you are using (gasoline) while simultaneously giving yourself the maximum flexibility (i.e. component parts) as well as the ability to change out those components without significantly interrupting the performance and/or functionality of the car? What changes do you suppose are required? A different fuel injector? Cold intake filter? Ha ha, no, that’s not nearly enough…not even by an order of magnitude.
In order to reach the expected gains in efficiency and flexibility, you will need to replace the piston engine with a gas turbine (the most efficient gasoline fueled power plant) and directly couple that to a DC generator and then install heavy magnet DC motors at each wheel…all controlled by a sophisticated vehicle management system that monitors and actively manages every aspect of the operation of the car. This is the way to take that Dart to the highest realms of performance and efficiency while also having the complete flexibility (elasticity) to replace or upgrade any individual component (the generator, the DC motors, etc.) without adversely affecting the rest of the system. For example; want higher speeds at the wheels (more benefit)? Put in higher horsepower DC motors…the gas turbine doesn’t need to be touched; it’s running a generator.
This is exactly what you do to your virtualized datacenter when you transform it into a cloud infrastructure. Just like in the Dart (by having the gas turbine isolated from the DC motors at the wheels), you need to separate the business benefit of the datacenter’s infrastructure from the operation of that infrastructure so that both can be individually optimized without significantly affecting the performance or operation of the other. This “decoupling” is done via automation and orchestration and by effectively abstracting all of the commodity activities (i.e. commissioning VMs, managing bandwidth and capacity, running multi-process operations, etc.) from the operations that actually provide value to the business.
That Guy’s Not Touching My
In the case of the now super-efficient and powerful Mega-Dart, it is the fastest, best handling, most fuel efficient version of itself you can possible create…but (99% of the importance of every conversation comes after that word)…the same guy who just a while ago was quite successfully working on your fuel injected, turbocharged Dart is now entirely unqualified to work on your Mega-Dart because when he looks under the hood, he sees nothing he recognizes. And, it’s not just a little different, it is so far removed from his experience and expertise it might as well be a nuclear submarine he’s looking at…so you don’t want that guy touching your car. It’s the same when you transform from a
traditional or even a virtualized datacenter to a cloud infrastructure; the new and entirely different category of technologies, components and processes that are a normal part of what makes it run are just like what’s under the hood of the Mega-Dart; vastly different from what is expected and wholly outside the experience and expertise of those who used to, quite successfully, engineer and operate the place before.
Different management and administration methodologies have to be utilized and different operational processes have to be devised…basically, different ways of thinking about how your business discovers, designs, creates, consumes and manages IT services have to be found. While there are process frameworks and models to help you (ITIL, CobiT, etc.) as well as organizations (GreenPages is a great example), ultimately—and this really is unavoidable—once you change the way you deliver IT services to your business via a cloud infrastructure, you will need to concurrently change the very essence of how you manage that infrastructure.
Imagine spending hundreds of thousands of dollars on transforming that ‘68 Dodge Dart into the Mega-Dart; the most efficient, fastest and flexible car conversion available, yet, it sits in your driveway gathering dust because no one knows how to operate it let alone turn it on.