Cloud Conference: Moving From Local to Cloud Infrastructure

ConductorIO VP of business development and operations Monique Bradshaw talked about the paradigm change from local, on-premise infrastructure to the cloud. “The paradigm shift means a fundamental change in approach of underlying assumption,” she said during an ETC Cloud Innovation Conference keynote at NAB. “We’re seeing a big change in the ways that companies are looking at their rendering.” In five years, she noted, 90 percent of respondents to a survey think they’ll have at least some of their rendering in the cloud, up from close to 60 percent today.

Bradshaw reports that the first big motivator driving change is cost savings. “The financial challenge of rendering in the VFX industry is that resources are actually over-provisioned during the workflow — sitting there, not used,” she said. “Then it goes through times when it’s under-provisioned, where an artist hopes his job will get on the farm. Companies hope to address this by transitioning to the cloud.”

ETC_Cloud_Puzzle

She also notes that big data is challenging to accumulate and store on premises. “When you leverage the capabilities of the cloud you have an infinite amount of storage and large database accessibility,” she said. “It exposes where your money is being spent and where your inefficiencies are.”

Scale and flexibility are two main reasons why people use the cloud. Bradshaw used Atomic Fiction’s work on “The Wire” to show the big dips and peaks of processor hours per month. “It would have cost $1.5 million to accommodate the resources needed in on-premise sources,” she said. With the cloud, the production enjoyed a cost savings of almost 50 percent ($750,000).

Making the transition to the cloud is tricky, she admits, but that’s where her company’s solution comes into play. “We have a central application that functions as the brain of the entire system,” she explained. “We aim to simplify it.”

Studios tend to take one of two approaches in transitioning to the cloud, she noted. “The first and most common is the hybrid model — local/cloud combination,” she said. “The benefit is that you utilize existing infrastructure and compute capacity. You get the fastest possible turnaround with the on-premise resources. You also maintain control over that local infrastructure. But when you hit that line you can then ‘burst’ into the cloud, which allows resource flexibility, pay for use, without a data center being necessary.”

The second methodology is to focus on cloud only, she said, which Atomic Fiction chose to take when they started out six years ago. “But workstations in the cloud aren’t really there yet,” she suggested. “So if you have workstations, you have render capacity on-premises that you don’t really want to leave idle. So everyone is in some form of transition in the hybrid model.”