Editor's Note: This story was originally published May 14, 2015.

The Federal Communications Commission has begun an IT Transformation project with the goal of eliminating all its legacy systems and moving to a public cloud infrastructure by the end of 2016. Leading this charge is CIO David Bray, who recently spoke with Federal Times Senior Writer Aaron Boyd about why an IT transformation — or intervention — is needed now. Below are edited excerpts from that interview.

What does the current infrastructure look like at the FCC?

When I arrived about a year and a half ago, I discovered that for an agency of only 1,750 people, which is fairly small, we happened to have about 207 different IT systems. Now, in that last year and a half we have managed to bring it down to about 102 systems with some rationalizations and consolidation. In terms of data centers, we actually have two data centers on-site in the building and our goal is by this summer to actually get out of the business of actually maintaining a data center.

Of those systems that I mentioned, more than 50 percent are more than 10 years old. And so in some cases it is actually where we cannot even find equipment or software that is continued to be supported for those old systems.

We actually want to move everything and basically do what we are calling "a lift to prepare for our shift." So the first phase of the lift is move everything to an offsite managed service provider so we can begin to rationalize what needs to go to the cloud and eventually get out of the business of maintaining any infrastructure with that managed service provider. Then the shift part is exactly that. How do we translate what used to be data that was actually stored in applications to accommodate a platform so we can age out of that business?

We have had some early successes. One is a new consumer complaint system that we have for the help desk. That was actually something where we had a requirement from both Congress and the administration to come up with a successor to the legacy IT help desk that we had for consumer complaints.

We were quoted it would probably take between a year to two years to do and about $3.2 million. We just did not have that time, let alone that money and thanks to doing software-as-a-service, we did it in less than six months and for about one-sixth the price — for about $450,000.

So that is the sort of thing we want to do as we do that lift and shift, moving things off-site and then to the cloud.

I think in the past what happened is whenever a new request came in — whether it was from the administration, or from Congress, or some other group — the bureaus and offices stood up a new IT system. And that worked well for the first five or 10 years but now we are at a point where we have got one system for every nine people.

I kind of feel like I am Oprah Winfrey — look under your chair, you are going home today with a system, everyone gets a system for free. And so it is just not sustainable. What we want to do is have a common data platform that is public cloud-based.

FCC is launching a new website in September — what have been some of the challenges there?

The website is a fun story of the FCC and I will sort of preface it with first the joy of any new CIO is sort of like buying a used car sight unseen, driving if off the lot and then opening it up and saying what did I just buy and looking underneath the engine. So the FCC's website has about 15 years of material on it. And it actually has multiple domains, some that date back as late as the mid to late 1990s and early 2000 period.

We are narrowing it down to two and we will probably know in the next months as to which one it is going to be. By September we expect to be fully rolling.

The next hard part though is actually taking that content and identifying what content needs to stay, what needs to be retired, what needs to be archived. Now, what we have been doing as a parallel process to get ready for that is we actually involved both stakeholders internally as well as some of the stakeholders externally because we realized that at the end of the day the FCC is really a business-to-business. I often say that my wife loves me, but she does not feel a compelling need to check the FCC website on a daily basis. It really is about the interactions that we have with different public stakeholders and different members of the private sector that the FCC does its business.

And so we have gotten their input as to how best to navigate that site because it is a lot of content. We actually used sort of eyeball tracking. So we actually had people who brought them in and they volunteered. We could track where in the pages their eyes were gravitating to so we knew how to do it better. It is the same sort A/B testing that you see Google, Facebook doing when they are actually trying to do web design.

We have a prototype and so we expect by September both to have the public cloud platform for the website plus a prototype actually based on how people are actually going to use the site so we can begin to migrate that content over. And over the next sort of six months rolling from there then actually have a site that actually is modern, it is on the cloud, it is much easier to do updates to but on top of it is actually something that someone is going to use.

How close are your major systems to actually being at end-of-life and what kind of clock are you on to make this transformation before you have got some serious issues?

The good news is we are not on Windows XP. We are on Windows 7 so that is good. The bigger challenge is in some cases there are platforms being used that I probably would not recommend because they are very close to end-of-life and so that is partly why we are doing the lift part, the moving the servers off-site this summer. And then in other cases I would say the clock is probably established to have a common data platform before the end of this year and begin to address a modular approach to some of these systems. Really hit the ground running in fiscal year '16.

We did ask for more money this year and we did not get more money — and so we are trying to do what we can with the funds we have which is always a fun challenge. But really we are hoping that the rocket ship takes off in fiscal year '16.

I have no doubt that these systems in some cases could continue to run for five or six years but what would happen during those five and six years would be the cost of maintaining them would go up by a matter of 10 or 15 percent each year. And we are already at the point where about 80 percent of our budget is being consumed by O&M, which is just not good. It is not sustainable.

While the systems could probably keep on running for four or five years even though they are already 15 or 20 years old. I sure would not want that to happen just because we would not have any fuel or funds to do what we need to do.

What's the plan for the IT Transformation program?

The goal really is three stages. One: move the servers off-site, which we will do this year. Two: establish that common data platform, which will be done by the end of this fiscal year. And then in '16 for licensing and tracking make sure that we are now moving to a process where we have that sort of service catalog of things that can be remixed. Because I have no doubt while licensing has commonalities there may be for AM [radio] you need to have these special data fields or this special process — versus satellite where you need this special process. So we allow for diversity and whatever the unique nuances that need to be done for different things. But at the same time we are not creating or we are not having to maintain 100 different systems for just one purpose.

So I expect by fiscal year '16 we will have demonstrated that we have established the common data platform. We have demonstrated we can make progress on licensing and tracking and then from there we can actually start to turn off old systems to fund anything else that we need to do in '17 and '18. So there is no new additional monies that are needed at that point.

We are not moving as fast as I would like because of what happened in '15 but that is how it goes. I do not get to determine how much fuel I get for the car.

Did you bring on any new people to help make this happen or you are doing it with your in-house staff?

Yes, there were some people let go so we could let in new people. So it is basically a net-net. And we needed to do that partly because some people had gotten used to a certain way of doing things. And their model was based on maintaining legacy systems and the reality is we do not need that model anymore.

But the other thing we needed was that new way of doing development so they can show other people how it is done. The other thing that I also need to do is balance running the trains on time. We do need the existing people for that who are really good about running the trains on time because I cannot forget that and at the same time set up for the next so that things can be inherited by those people that make the trains run on time.

What's the timetable for seeing a return on investment?

We have some milestones identified that there is going to be at least a 30 percent reduction in the amount that would be required to maintain going forward. And so if you figure that we spend on O&M — so this is maintaining the things we have, no new development — we probably spend about $55 million. If you see a 30 percent reduction in that within the first two years, if not first year, you will actually see the return on investment. It is just a question of do you do that bump up at beginning. So it takes money to make money in some respect or to save money.

What kind of risk are you looking at specific to this modernization effort?

There is the risk of course of obviously trying to do something because it is something that has not been done before. We are trying to follow the agile development model, which means we are experimenting and we are seeing what works and then carrying it over. But at the same time there is the risk of doing nothing and so you have an event where a system that is old, more than 15 years old suddenly has an increased demand like we saw in the summer and then that system just cannot meet that increased demand because it was built for the 1990s as opposed to now.

So it is really the risk of doing nothing, which I think actually exceeds the risk of trying to do something. And then just trying to figure out how we explain both to the bureaus and offices what we are doing? How do we explain to the public what we are doing? How do we explain to our stakeholders? I think there is a risk that we can move too fast. But at the same time there is also risk we can move too slow.

Is the IT Transformation a program with an end date or more of a philosophy?

It is an intervention. The way of maintaining IT on-site and building systems that had a very long O&M tail that we are just multiplying, in some respects like rabbits, just can no longer work. It really is an intervention because the way things were working could not just be tweaked. It actually had to be sort of a transformation of how we were doing things. That is why we are calling it the IT Transformation.

And so by moving the servers off-site that is the point of no return. By setting up the common data platform and saying all new development will be using the new common data platform. That is a philosophy embedded within the intervention. And so while you are right there is an end date to what we hope to be in terms of the intervention, we hope that the intervention introduces this whole new set of develop philosophies — ways of doing IT across the FCC that will continue long after I leave or the program ends.