Endframe Case Study

Endframe needed a scalable, high-end production solution for virtual events that allowed them the flexibility to allocate resources for different sized events, while also enabling their crew to work from anywhere in the world; regardless of the COVID restrictions that may have been in place.

Endframe chose to engage with Fitzroy IT because they were uniquely qualified to take on the challenges presented by cloud-based video broadcasting. The project required broad experience across AWS Infrastructure Management, Systems Administration and Video Production. Such diverse skills are rarely found in one place, but Fitzroy IT prides itself on the breadth and depth of its knowledge and expertise.

Endframe were already using AWS Media Services for their streaming pipeline, so moving the rest of their production to AWS with their EC2 G4dn NVIDIA GPU instances was the best option. Choosing AWS was a straightforward business and technology decision; guided by Fitzroy IT, Endframe could see that no other provider could meet their needs for bandwidth, performance and high-availability at an extremely competitive price.

Initially, Fitzroy IT worked with Endframe to understand their complex requirements and design a new AWS infrastructure.

As a high end production company, Endframe were using a number of different video software applications to meet the requirements of the varied events and clients they work with. These included Zoom, vMix, NDI Tools, Premiere Pro, Vicreo Listener and Office 365. This software needed to work seamlessly in AWS and be accessible via remote technicians using multi-monitor setups.

Fitzroy IT took all of these disparate parts and created AMIs and setup scripts to automate set-up. The deployed solution runs on G4dn EC2 instances, utilising the NVIDIA GPUs to accelerate all intensive video tasks. The EC2s communicate with S3 and AWS Media Services. Users connect to a Client VPN to securely access the infrastructure, and then use Nice DCV for a Remote Desktop experience with multiple monitors.

Rolling out this solution with AWS has improved reliability, saved money, and provided scalability with increased security.

The new AWS infrastructure was built up in just under three weeks and it was first used for a large Australian scientific organisation.Three, two-day conferences were run, each supporting up to two concurrent video streams, 40 hours of streaming sessions, 700 attendees and facilitating workgroup activities with up to 10 breakout sessions.

The infrastructure allows one or two technicians to manage all of those sessions via remote desktop, and only spin up those instances an hour before the session begins to minimise costs.

While the current implementation is doing everything Endframe needs, there are some exciting capability enhancements possible. These include:

• Developing an Infrastructure-as-Code version, using AWS Cloudformation
• Creating a PaaS version that can be leased out to other broadcast organisations
• Enabling Multiple Availability Zones for redundancy and access to additional GPUs