Now that Williams and TWC have had one success with the cloud, they're looking at it for other applications. "We're really, really excited about the opportunities the cloud can bring us," he said at AWS's recent user conference, which he attended to learn more about the products and services offered from AWS's cloud and its partner network.
NEW VENTURE: EMC and VMware spin out big data/cloud division
Another potential cloud use for TWC is having access to compute resources as developers need them, allowing an application or system to be developed in-house and then tested almost immediately across the cloud. "It'll allow us to react to our business better," Williams says.
TWC has spent the past few years migrating to a largely virtualized environment. The cloud, Williams says, is the next logical step. "There's a lot of POC [proof of concept]," he says. "Everyone sees the value, but we don't want to do it because it's the cool, trendy thing to do; we want to do it to add to the experience of our users."
There are a lot of considerations to take into account when moving from the multiple data centers the company currently runs most of its programs out of. No. 1 is making sure the system works. Amazon Web Services, which is where Williams and his team have dabbled the most in using cloud resources, hasn't been immune to weather events itself. This past summer's severe storms in Northern Virginia caused a power outage that led to AWS's cloud being unavailable for some customers. TWC, naturally, needs to make sure its network is up during such times.
That's why fault tolerant and highly available architectures are top-of-mind for Williams. He hopes to deploy Chaos Monkey, Netflix's randomly generating fault tolerant testing system, to test the cloud architecture for failures. Chaos Monkey is open source software programmed to randomly shut down parts of a system to test if the application is fault tolerant.
Another big part of having a highly available system, Williams says, is not being tied to a single vendor. The goal is to have no single point of failure, he says.