Dark architecture addresses some of the issues, and lessens the risks, for companies that are switching over and making major changes in their infrastructure.

When developers and businesses have to change their existing IT architecture, such as scaling up new features or swapping out an old legacy system, it can be a risky proposition.

This is where Dyn’s dark architecture approach to upgrades comes in, according to Dave Connors, VP of technical operations at Dyn, speaking at GigaOM Structure. When you have a legacy system, changing features could could require nine to 12 months of effort that requires a lot of time and developers. And once projects are completed and the IT organization flips a switch, the new, improved system turns on.

The immediate switch can lead to sleepless nights for developers and CIOs and is risky for the business that depends on the system functioning to operate. Plus, it will take almsot a year to get any value from an IT upgrade. That’s too risky and too slow.

Dyn has build an approach to lessen these risks and deliver some value sooner. Connors calls it the company’s “dark architecture” approach. Under this approach, it breaks the project up into small pieces that focus on the most painful business problem. Dyn looks at an IT process as a series of flows with the legacy black box in the middle, with traffic flowing in and traffic flowing out.

So you can look at the problem as a system of flows: you have the black box in the middle, with traffic going in and traffic going out. Then you construct those flows and identify the top priorities that need fixing. Once you have done that, you pick the top priority and set up a parallel system of flows that mimic the inputs and outputs.

That particular flow might only require 2 percent of the entire problem, but it’s a huge pain point. So you identify the steps in that flow and change your parallel input and then see what happens on the output side. IT can monitor the results of the change for however long it needs to in order to ensure that the flow is working like it should. But the user isn’t receiveing it. Then when IT is sure the system works, it throws the old output flow away and the most painful aspect of the old way of doing things has been solved.

Yes, there’s still 98 percent of the old system left, but customers are experiencing the value of the shift sooner than they might have under the old forklift-style of upgrading systems. IT can do this for as long as it is solving the critical problems until the entire legacy system has converted. The customer bears the cost of running the two systems in parallel, but cuts down on the risk of a massive deployment or spending a year on an IT project that is irrelevant before it is even finished.

“Dark architecture addresses the two major challenges of infrastructure upgrades,” he said.

Check out the rest of our Structure 2013 live coverage here, and a video embed of the session follows below:

A transcription of the video follows on the next page

page of 2

Comments have been disabled for this post