It's easy to put off DevOps as just another trend: Culture change is hard. But your competitors aren't waiting.
Should you consider a chaotic architecture?
Recently the Jet Propulsion Lab’s technology leadership began a collaborative conversation about chaotic architecture as a cornerstone of the next IT decade (that is, 2015-18). As CIO, Jim Rinaldi, and IT Chief Technology Officer, Tom Soderstrom, define it, chaotic architecture at its core means building technology tools based on changeable modules with a short half-life and securing data while viewing applications as temporary. The Enterprisers Project caught up with the two at their offices in Pasadena, CA, to learn more about this fascinating concept, and how to determine whether it’s a good fit for your enterprise.
The Enterprisers Project (TEP): Who formed the idea of a chaotic architecture?
Soderstrom: Jim created it, but it may not be fully baked yet.
Rinaldi: We are creating and living it. For example, I’ve done a few talks where there are enterprise architects in the room, and they come up to me and say, “I want to be a chaotic architect, not an enterprise architect.” What they’re really saying is that building in a flexible architecture and having options makes more sense than trying to have a very rigid architecture that becomes unsupportable within a few years.
TEP: And yet the name sounds a little counter-intuitive since chaos is about unpredictability but also lack of order. Not what you’d typically want in an IT department.
Rinaldi: Years ago we tried "enterprise architecture" here at JPL, and no one was really resonating with that. We had so many different architectures, so many different ecosystems, that it became difficult for us to explain what we were trying to do using that term. It also didn’t represent what we were trying to do in reality, working in an R&D organization.
With the way things are evolving and shifting in IT, with so many choices and opportunities to solve many problems differently, it seemed as if chaotic architecture might be more applicable to what we’re living and planning for now. My guess is if you ask most companies what their enterprise architecture is, they couldn’t tell you outside of a few people in IT.
Soderstrom: We asked ourselves whether we wanted to fight an IT reality of "Let’s build it to be changed" or just build that reality into the fabric of our culture.
TEP: So it wasn’t something you developed as a model for one product or project.
Rinaldi: No, it’s for everything. It also might help to back up a bit. At JPL we’ve been working to become a consulting organization as well as a services organization. When you consult, you want to help enable non-IT organizations to be the most productive and effective they can, but not necessarily the most efficient. And one of the things that we noticed is that our tool sets and services were not seen as very competitive in our customer’s mind.
One example is storage. We assumed that we had to provide the best, most elaborate storage known to mankind. And it turns out our customers were going out and buying cheap storage, storage they could afford, and that could break. So we ended up buying that kind of storage and put it into a capability that some of our teams could use instead of going out and buying it on their own. That taught me that our IT group might have been pricing ourselves out of opportunities and that as a consulting organization maybe we weren’t really listening to our customers’ demands in a consistent way.
Soderstrom: That’s a great back-end example. On the front end, we support a lot of tools for collaboration: WebEx, Lync, Slack, and HipChat. If we said, "You must use one way to communicate," it would never work. We look instead at what they are using and whether we can enable it. That’s the chaos part. The only "institutional" product we provide in this area is Jabber.
TEP: Wearing my traditional CIO hat for a moment, I have to ask if the architecture is chaotic, then how is it managed and how is it secured?
Rinaldi: We have a core set of things in the architecture that we in IT and the broader institution can rely on, such as our financial system, our network, and email. But we’re also aware of and support other things that become more individualistic or team-oriented that may play outside of the core. The fact is that we’re a multi-technology company. We have Microsoft, we have Apple, we have Linux, and we have a lot of variations of Linux. We have a pretty wide adoption of open software, because the new generation likes to work with open software. People tend to use the technologies that help them do their daily jobs.
But what we figured out is trying to make these environments work together as we’d like just isn’t going to happen. In an R&D environment the skill sets we bring in and the pedigree of the personnel we hire are pretty high. And we get the benefit out of the results of the research and engineering they do by giving them flexibility versus trying to force them to use our IT.
You also brought up the idea of security, which is first and foremost for us. We have security plans across all our environments, as well as the monitoring and technology capability to not only look at what’s going on in the network but what leaves the network. Certainly we know what to block coming into the network. We’ve taken a risk-based approach in protecting the crown jewels and making sure that we have smart data about our environment such that we can do analytics on it and ultimately get more insight versus throwing human labor at it.
Soderstrom: One of the things that we realized as part of this process is that applications are temporary and data is permanent. So if we think "data first" and make sure that data is protected and can be reused, then however they get to that data, end users can write their own applications. Which is part of a chaotic architecture. Users and not just applications groups will write their own applications.
Rinaldi: And the result in a chaotic architecture is that IT owns the architecture but makes it more flexible for, in our case, our laboratory to benefit from.
TEP: You’ve mentioned a few times that JPL is an R&D organization. So I have to ask, would a chaotic architecture work for every kind of enterprise?
Rinaldi: Probably not. But it is a trend that we’re seeing. I talk to a lot of CIOs, and I don’t think this is one of those things that will go away if you close your eyes. The Internet of Things, big data technologies, all those things that are coming our way have shown us that our worlds are going to get more exciting. And if you try to shut it down, avoid it, or block it, that’s where IT gets minimized and other IT gets created.
Soderstrom: And to play on that word exciting, if we look at the future, and how we’re going to attract that next generation of IT talent, an old architecture with old tools will not get anyone excited to work here. But if we do make it exciting and have an architecture that has built in all the layers of tools people need, they will come in and hit the ground running immediately with those new tools.
TEP: A chaotic architecture is just one part of JPL’s next IT "decade," which will end in 2018. If this chaotic architecture continues to come to fruition the way you think it will, how will the world be different?
Rinaldi: From my vantage point we want our base of smart, intelligent users to have different ways of working where it’s more seamless and less labor-intensive for them to get to the information and data that they need. Those users deserve to have choices in accessing what they need; not only from a technology standpoint but based on how they are able to leverage prior knowledge here at JPL.
Soderstrom: The only thing I would add is that we hope to have more user participation. We already have a lot of smart people at JPL, but can we get them to participate directly into the mission? For example, if we want Google-scale internal search, we teach people how to write search connectors. It saves money. It saves time. And if they came up with the idea and they can implement the idea because it fits with the architecture, everybody wins.
Tom Soderstrom serves as the IT Chief Technology Officer at NASA's Jet Propulsion Laboratory, where his mission is to identify and infuse new IT technologies into JPL's environment. He has led remote teams and large scale IT best practices development and change efforts in both small startup and large commercial companies, in international venues, and in the US Government arena. Some of the companies he has worked for include Telos, enterWorks, User Technology Associates, Digital Island, Exodus, Cable & Wireless, and Raytheon.
Jim Rinaldi is the CIO of at NASA's Jet Propulsion Laboratory.