The secret to building primary storage for the private cloud

As with all things “cloud” these days, there’s an interesting dichotomy between those who are head over heels excited about the huge potential of private cloud approaches and those who see it as a marketing-driven rebrand of what everyone is already doing: virtualizing the data center. In the end, the truth is somewhere in between.

Although the private cloud has the potential to solve many of the scalability and service-delivery challenges that larger IT departments wrestle with, it is not magic. Really reaping the benefits of the private cloud requires an IT-savvy customer base, a willingness from the executive suite to adopt a chargeback model, and no small amount of work to configure it in such a way that it will actually save everyone time in the end.

At the same time, a true private cloud implementation is not just data-center virtualization with a new coat of paint. Although it’s true that much of the gear you might use in a private cloud ends up being put together in the same way as you would in a traditional virtualization model, the self-service nature of the private cloud demands a completely different mindset from those planning and building the underlying virtualization and storage infrastructures.

What’s different about private cloud storage

In a typical IT department, planning and budgeting for new infrastructure expansion usually happens many months before the beginning of the fiscal year and is based on the perceived demand from business units for various data center resources — whether they be compute, network, or storage. However, even though that process is becoming harder and harder to do accurately — especially as storage demand grows at ever-increasing rates and becomes more difficult to predict — that’s still the way most of us do it.

In the end, IT is ultimately in the driver’s seat: deciding what infrastructure expansions are made and when. When a business unit inevitably discovers that it needs to dramatically increase its resource consumption (say, to field a new application), a usually lengthy process commences of figuring out where the money will come from. Finally, after a few months of wrangling, the service is delivered and life goes on.

Therein lies one of the key differences between a traditional data center and a private cloud infrastructure. Due to the self-service nature of the private cloud, you may have literally no warning whatsoever when an enormous usage spike might occur. Thus, you need to maintain enough slack capacity to absorb a reasonable amount of growth and be able to react extremely quickly to satisfy demands that outstrip your available capacity — much in the same way that a typical service provider needs to.

Staying ahead of that kind of unpredictable growth demands an infrastructure that is designed to be grown in small increments under very short timetables, as well as the implementation of monitoring tools that can track and trend the consumption rate of available resources.

The pitfalls of delivering storage in the private cloud

Those challenges are certainly not insurmountable, but the solutions come with their own pitfalls. For example, budgeted IT infrastructure investments are typically a fairly substantial projects that might implement an entire year’s worth of expected storage growth at once. As anyone who has shopped for enterprise storage knows, vendors are typically much happier to give you a steep discount on a few hundred thousand dollars of investment executed at all once than they are for a few tens of thousands of dollars spent here and there throughout the year.

Solving that challenge usually requires negotiating with the vendor to get a fixed discount and accurate lead-time expectations on storage gear for the entire year. That way, you can match your chargeback rates to your underlying costs — whether you’re buying a whole new SAN or just a few disks.

The demand for that sort of incremental growth may also sway your thought process when it comes to what kind of storage gear you opt to use. Some storage hardware is simply impossible to grow in small increments, whereas others may scale gracefully up to a point where an expensive and disruptive forklift upgrade is required. Fortunately, storage vendors have recognized this challenge. Some are increasingly delivering scale-out storage gear that can grow linearly. Others are focusing on built-in storage virtualization layers that cluster multiple scale-up storage infrastructures to achieve similar results. Both approaches try to avoid the pain of the forklift upgrade.

Putting it all together

No matter what gear or monitoring tools you use, it’s important to recognize that life after the private cloud will be very different than life before it.

Once you make the adjustment, you’ll be better off. Having the freedom to expand the infrastructure as demand requires it rather than struggling to plan accurately years in advance is actually a benefit, not a curse. But it does require a different approach to overall design and to how you buy your gear — both good to know before it’s too late.

This article, “The secret to building primary storage for the private cloud,” originally appeared at InfoWorld.com. Read more of Matt Prigge’s Information Overload blog and follow the latest developments in storage at InfoWorld.com. For the latest business technology news, follow InfoWorld.com on Twitter.