Setting the Record Straight on Our Cloud Strategy

By Chris Rice, AT&T

If it feels like a lot of our focus has been up in the clouds recently, you wouldn’t be wrong. Last month, we announced our intention to move AT&T Business applications to IBM Cloud. A week later, we made Microsoft our preferred cloud provider for non-network applications. And just last week, we highlighted a new collaboration with Dellto accelerate the deployment of our AT&T Network Cloud by enhancing common infrastructure functionality through open source communities.

On the surface, all of this talk might seem a bit, well, cloudy, but in reality, our cloud strategy has never been clearer.

Today, we’re using cloud technologies to modernize and streamline our business applications, accelerate innovation and new services with customers, and quickly deploy software-defined network (SDN) technologies and virtual network functions to power our own network.

All this work isn’t happening in the same cloud, though. To better understand our approach, it’s helpful to understand the difference between what kinds of workloads and applications we’ll run in each cloud, and how and why we utilize them.

Our 5G Network Lives in Our Network Cloud

We started on a path for a single cloud, called AT&T Integrated Cloud (AIC). This was our private cloud, meaning we managed all the workloads and infrastructure within it. Originally, AIC housed both our network and several of our “non-network” IT workloads and applications.

But we quickly learned it wasn’t optimal to combine both types of workloads on a single cloud. It required too many compromises, and the IT and network workloads needed different profiles of compute, network and storage.

We opted for a better approach: Create a private cloud for our network workloads, optimize it for those workloads, and drive the software definition and virtualization of our network through this cloud approach and through the use of white boxes for specific switching and routing functions.

Last year, we introduced Network Cloud as our next-generation cloud platform. As its name implies, it’s optimized specifically for AT&T network workloads. It’s powered by open source software running on commodity and white box hardware. That lets us keep costs down and innovate at a speed unmatched in the industry.

This is the path we’re on now, and the one aligned with our recent cloud announcements.   

At the foundation of our Network Cloud is an under-cloud platform called Airship, an open infrastructure project we jointly launched last year with SKT, Intel and the OpenStack Foundation. While there’s a lot more to this platform than I’ll go into here, Airship uses containerized software in a declarative way to automate what used to be the mostly manual process of building, managing and upgrading our cloud.

What does that mean? In short, that automation also lets us deploy, manage and upgrade our software-defined network, as well as the workloads that run on it, more quickly and seamlessly than we could in AIC – think days rather than weeks and minutes rather than hours, all without major service interruptions. And because it’s open source, it’s a platform others in the industry can take advantage of, too.

That’s why our Network Cloud was the perfect environment to house our new mobile 5G packet core. In fact, our commercial 5G network was the first network born in the cloud.

And while our internal teams are hard at work bringing 5G capabilities into the market at break-neck speed, open source collaborations with technology leaders like Dell and others will continue playing a vital role in our success as well. The more we work with others to fine-tune open source platforms like Airship, the faster we can improve our Network Cloud to quickly deliver new functions and capabilities into our 5G network.

But we’re also becoming a public cloud-first company.

We’re turning to public cloud providers to host our non-network workloads. Think traditional IT applications like billing and customer care, and corporate applications like HR and finance. By 2024, most of these cloud workloads will be housed in the public cloud

Moving these applications and workloads to the public cloud allows us to take advantage of the elasticity and cloud economics while letting us focus on what we as a company do best: deploying and running world-class networks. We can empower our employees to create and bring new, innovative network services and products to our customers while lowering costs. We can deliver a faster, more reliable, more responsive and more agile network, both in the core and at the edge. And, we can do so while simultaneously optimizing IT systems necessary to run it.  

These shifts are already paying off.

Last December, we became the first U.S. carrier to launch standards-based 5G service. And we’re well on our way to bringing our commercial 5G network nationwide by 2020.

While impressive in their own right, those feats wouldn’t be possible had we not recognized early on that when it comes to cloud, a hybrid approach lets us put the right workloads on the right clouds. That’s a model we know works, and one we’ll continue leading to deliver a best-in-class network for our customers.