ACADIA Blog

Controlling the transition to the cloud: lessons from the 90’s

The rush to the cloud has begun. That could be a good thing—but only if we take the time to learn from the lessons of the past.

Think back to the x86 proliferation in the mid-1990s. With x86 machines suddenly more affordable and robust, and with Internet use becoming widespread, consumers saw a chance to speed their time to market. Why wait 12-18 months for central IT to get a new application up and running, they figured, when they could do it themselves in a matter of weeks?

Naturally, they went for it. The result was massive IT disruption: sprawling, unmanageable infrastructures, insecure data, information silos. It took years, and a lot of money, to try to sort it all out.

Fast forward to today. With the advent of cloud computing, we are on the cusp of another extensive IT transformation. Once again, consumers of IT resources have the opportunity to get what they need cheaper, faster, and with more control. With the public cloud, they can provision a server in minutes and pay for only what they consume. Naturally, they’ll want to go for it.

This actually could be good for everyone: most of the workloads running in the public cloud are the kind CIO’s should be glad to get rid of. If the big enticement to go to public cloud is the pay-as-you-go format, it follows that most of these are on-again, off-again workloads—in other words, they are unpredictable and volatile. These are two words that don’t fit into the clean definitions we like to put around service level guarantees. CIOs should look forward to off-loading them.

The problem is that there are some workloads that shouldn’t leave the premises—data that contains intellectual property, that needs to be protected for regulatory and compliance reasons, or that contains private information. If consumers are free to shift everything to public cloud, without policies and audit processes to guide them, the disruption will be costly.

The cloud is coming and, just as they did in the mid 1990’s, the consumers of IT resources will force the issue. But we don’t have to re-live the mistakes of the past. There are steps you can take today to prepare your organization for a safe, orderly, and effective transition to this new IT paradigm.

More on that—and on the importance of bringing everyone into the conversation—in my next post.

Your CIO is going to ask you to manage thousands of VMs without adding staff. Can you do it?

If you can see a light at the end of the tunnel right now, it probably looks a lot like an oncoming train. While the recession created a pent-up demand for new applications and a big backlog of needed technology refreshes, it did nothing to fatten IT budgets.

Which leaves you with a big problem: how do you satisfy your customers’ need for change with the staff and budget you have?

Reaping the OPEX advantages of virtualization

Imagine what you could do if it were possible for a single, mid-level IT professional with general skills to manage the equivalent of a mid-sized data center—1,000 virtual machines, a 10-gigabit network, and 40 terabytes of storage.

This kind of super efficiency is actually a lot closer to becoming reality than you might think. The potential OPEX efficiencies of virtualization are huge—much more compelling, in fact, than the CAPEX cost savings everyone understands. Yet very few organizations are reaping any significant OPEX savings from virtualization today beyond simple provisioning. Equally important, very few have the processes, procedures, and tools in place to manage their increasingly dense, virtual, and converged infrastructures.

The solution to both these challenges, I believe, is to change the management paradigm—from the operation of infrastructure to the orchestration of people, processes, and technology.

The end of the tunnel?

In many ways, managing a large-scale virtual environment has more in common with conducting a successful orchestra performance than with our traditional infrastructure management practices.

Rather than dividing up the work and performing largely physical management tasks in separate, specialized infrastructure towers, organizations need a way to build workflows that bring the right virtual elements together, in the right way and at the right time, to achieve a specific result.

Orchestration is a methodology for doing that. By bringing together the right policies, people, processes, and technologies in the right workflow to resolve an event or satisfy a request, orchestration makes it possible for one professional to manage what it usually takes multiple staff to handle.

The good news is that the methodologies, tools, and services needed to achieve orchestration are already available. It’s important to remember, however, that orchestration is much more than automation or a package of new management software. But more on that—and other cloud misconceptions—later.

To learn more about orchestration, and to see how an orchestrated firmware upgrade on a switch compares with the conventional approach, check out my white paper, “Orchestration: A New Model for Operational Efficiency in a Virtual Environment”.