At many visual effects studios, when deadlines on multiple feature films and commercials compete for rendering time, demand clashes with capacity. Rendering workloads at studios like Framestore dramatically spike as deadlines approach. On occasion, half of the rendering cycles for a film are completed in the final month before the film’s release. Studios are then forced to purchase hundreds or even thousands of additional servers to accommodate this peak provisioning problem. After the productions are delivered successfully, studios switch off the servers to save power until they are needed again.
Compute Engine along with Cloud Storage and Cloud Networking form the foundation to help studios attack even the most daunting of rendering and resource-intensive image processing tasks. The big challenge for VFX studios when doing cloud-based rendering is that the input data resides on-prem at the studio’s data centers (or in a remote CoLo). With storage requirements for big productions today exceeding hundreds of terabytes, the physical distance between where the raw data resides and Google’s data centers can have a large latency impact.
Avere Virtual FXT Edge filers running as a cluster of virtual machines on Compute Engine offer a compelling solution for studios’ deadline-driven rendering work. The virtual cluster acts as a cache – effectively NAS in the cloud – and enables rendering nodes in Compute Engine to crunch file-based data that is already “cloud-bound,” avoiding costly latency. Efficiency gains abound given the nature of rendering workloads: files are cached on-demand onto high performance SSD storage, and can be served repeatedly to different rendering nodes. A high amount of input file reuse across frames is common, so a “hot” working set of cached files can often be utilized by multiple render nodes in parallel. The benefits of using Avere’s Edge filers extend throughout a studio’s production process beyond just rendering. Digital assets used throughout the entire production can be cached once and served thousands or millions of times to Compute Engine VMs without having to pull from colder storage sources.
With per-minute billing, the economics are more transparent and often more favorable compared to building and maintaining your own monolithic data centers. Resource use is purely OPEX (operational expense), on-demand with no waste or overhead. With preemptible virtual machine instances from Google, you can save up to 70% off the normal price for workloads where this makes sense. For example, if a backlog of low-priority jobs exists in your render queue, they can be cleared by tapping preemptible instances during off-peak hours.
Framestore has begun piloting Avere to support rendering on Google Cloud Platform.
"Just like how we used Avere for caching, we are now looking to see if we can use Avere to cache data [on Google Cloud Platform] so it does not have to take the long trip back to our servers," said Steve MacPherson, CTO at Framestore, in an interview with Computer Weekly. Moving workloads to the public cloud frees up rendering resources on-premises and could potentially save the company £200,000 in the cost of buying extra server cores.
Framestore deploying Avere’s technology on Cloud Platform is an excellent showcase of how a world-class creative studio is augmenting its existing operations with Google’s infrastructure. On-prem and cloud-based processing are not “either-or”: cloud-based rendering can help overcome real challenges faced by media companies with intense CPU and I/O workloads, freeing up resources locally. Ultimately, the goal is to enable inspiring companies like Framestore to focus on creativity, delight audiences and grow their business.
We invite you to join our webinar, titled Build a Cloud Render-Ready Infrastructure, on September 9th with Framestore, Avere Systems, and Google. Event details:
September 9, 2015
11:00 AM Eastern, 8:00 AM Pacific, 4:00 PM United Kingdom
Attend this Webinar
Until then, you can learn more about Avere on Google Cloud Platform or contact us to explore how we can tailor Google’s cloud to your operations.
-Posted by Andy Tzou, Product Marketing Manager, Google Cloud Platform
Compute Engine along with Cloud Storage and Cloud Networking form the foundation to help studios attack even the most daunting of rendering and resource-intensive image processing tasks. The big challenge for VFX studios when doing cloud-based rendering is that the input data resides on-prem at the studio’s data centers (or in a remote CoLo). With storage requirements for big productions today exceeding hundreds of terabytes, the physical distance between where the raw data resides and Google’s data centers can have a large latency impact.
Avere Virtual FXT Edge filers running as a cluster of virtual machines on Compute Engine offer a compelling solution for studios’ deadline-driven rendering work. The virtual cluster acts as a cache – effectively NAS in the cloud – and enables rendering nodes in Compute Engine to crunch file-based data that is already “cloud-bound,” avoiding costly latency. Efficiency gains abound given the nature of rendering workloads: files are cached on-demand onto high performance SSD storage, and can be served repeatedly to different rendering nodes. A high amount of input file reuse across frames is common, so a “hot” working set of cached files can often be utilized by multiple render nodes in parallel. The benefits of using Avere’s Edge filers extend throughout a studio’s production process beyond just rendering. Digital assets used throughout the entire production can be cached once and served thousands or millions of times to Compute Engine VMs without having to pull from colder storage sources.
With per-minute billing, the economics are more transparent and often more favorable compared to building and maintaining your own monolithic data centers. Resource use is purely OPEX (operational expense), on-demand with no waste or overhead. With preemptible virtual machine instances from Google, you can save up to 70% off the normal price for workloads where this makes sense. For example, if a backlog of low-priority jobs exists in your render queue, they can be cleared by tapping preemptible instances during off-peak hours.
Framestore has begun piloting Avere to support rendering on Google Cloud Platform.
"Just like how we used Avere for caching, we are now looking to see if we can use Avere to cache data [on Google Cloud Platform] so it does not have to take the long trip back to our servers," said Steve MacPherson, CTO at Framestore, in an interview with Computer Weekly. Moving workloads to the public cloud frees up rendering resources on-premises and could potentially save the company £200,000 in the cost of buying extra server cores.
Framestore deploying Avere’s technology on Cloud Platform is an excellent showcase of how a world-class creative studio is augmenting its existing operations with Google’s infrastructure. On-prem and cloud-based processing are not “either-or”: cloud-based rendering can help overcome real challenges faced by media companies with intense CPU and I/O workloads, freeing up resources locally. Ultimately, the goal is to enable inspiring companies like Framestore to focus on creativity, delight audiences and grow their business.
We invite you to join our webinar, titled Build a Cloud Render-Ready Infrastructure, on September 9th with Framestore, Avere Systems, and Google. Event details:
September 9, 2015
11:00 AM Eastern, 8:00 AM Pacific, 4:00 PM United Kingdom
Attend this Webinar
Until then, you can learn more about Avere on Google Cloud Platform or contact us to explore how we can tailor Google’s cloud to your operations.
-Posted by Andy Tzou, Product Marketing Manager, Google Cloud Platform
