X
Business

Serverless but not stress-free: enterprise computing moves outside the enterprise

Panel of CA distinguished engineers discuss the promise and perils of cloud, containers, serverless computing and artificial intelligence
Written by Joe McKendrick, Contributing Writer

Within the next 10 years, much of the IT infrastructure as we know it will be out in the cloud - yet it will be relatively commonplace to be moving applications back in-house, or between clouds. However, all this new agility will be tempered by the stress of needing to get applications out the door almost minute by minute.

ca-world-distinguished-engineer-panel-cropped-11-16-2017.jpg

L to R: Otto Berkes, Richard Philyaw, Debra Danielson, Howard Abrams, Craig Vosburgh

Photo: Joe McKendrick

These challenges and opportunities in tomorrow's enterprise architecture were discussed by a panel of CA's technology elite - distinguished engineers who shared their visions and concerns at the recent CA World event in Las Vegas. Otto Berkes, chief technology officer for CA Technologies and moderator, asked for the engineers' visions about the road ahead for enterprise computing. (Note: the panel discussion portion starts at the 60-minute mark in the recording.)

Richard Philyaw, distinguished engineer and senior VP at CA, sees accelerating adoption of emerging modes of enterprise computing, such as microservices, containers and serverless computing. Looking 10 years down the road, it's likely that much of enterprise computing will be outside the enterprise. "I think the best way to think about the next 10 years is to look back to the previous 10 years and look at what's changed to get us to this point," he explained. "It was just over 10 years ago that Amazon Web Services was released to the public, and now entire businesses are being run on AWS. So I think if we roll forward from here we'll see a point where many enterprises will no longer own their infrastructure at all."

Containers are paving the way to a more fluid type of architecture that enables more shifting of assets between on-premises and cloud environments. "We've moved from monoliths into macro services, and now we're moving to microservices and from microservices to a serverless environment," said Craig Vosburgh, distinguished engineer and senior VP of software development at CA, and formerly a flight controller for the NASA Space Shuttle. The emergence of tools and platforms such as Kubernetes, Swarm and Compose are helping to orchestrate a "new world order where we are scaling from three units, up to 100 units, and back down to three units."

The challenge for software developers these days is the need for speed -- lots of speed. Development teams are "struggling to go faster, they don't want to deliver once a quarter, or once every month -- they want to deliver every five minutes," said Howard Abrams, distinguished engineer and senior VP of engineering at CA. "How do they balance that and still test to make sure their software's secure and reliable, and don't introduce a tons of bugs? How do they balance all that and not build huge mountains of technical debt that they'll never climb?"

Complexity won't go away, either, no matter how much of the infrastructure is move to cloud. "You're going to end up with more and more of a PaaS layer that you're going be able to depend on," said Vosburgh. "But the downside to that is the increase in complexity -- we've moved from having one thing to having tens of things to having hundreds of things, and we're certain to move soon to move to having tens of thousands of things that we're having to manage." Monitoring these various components is also getting more complicated, he added.

Panelists also discussed the growing role of artificial intelligence in enterprise shops. In the rush to AI, many organizations risk adopting processes that work against privacy and security. "There's a dark side to AI and we're going be hearing a lot about this in the future," said Debra Danielson, distinguished engineer and senior VP for CA. "The core issue here is that we're learning in our machine learning systems to recognize patterns outside of historical data, and the problem is some of those patterns contain bias and bad behavior and we're at risk of having that bad behavior and bias driven and magnified into our systems. We really are going have to watch out for this and make sure that we don't allow our AI to be driven by our history and not by our values."

(Disclosure: CA assisted in my travel expenses to CA World.)

Editorial standards