AWS for M&E Blog
New multi-partner live cloud production workshop supported by AWS
As live cloud production (LCP) workloads gain more momentum, customers want to deploy, integrate, and scale their systems more easily while gaining workflow observability. This includes gaining insights into the infrastructure, exposing meaningful application metrics, and monitoring their media assets in the cloud with comparable operational familiarity to their on-premises deployments.
To help partners and customers accelerate LCP adoption in the cloud, HAQM Web Services (AWS) continues to host interoperability workshops that address various technical topics. We connect with our partners in the spirit of “break things, discover, and connect”, while pushing the limits of various system components. For example, we addressed the topic of resiliency in our January 2024’s workshop.
Background
In March of this year, we held another multi-partner workshop in partnership with the HAQM Prime Video team at the HAQM Studios in Culver City, LA. This location fostered collaboration with the Prime Video Sports Operations team who provided access to their facility’s broadcast infrastructure.
The event featured 13 AWS Partners including two System Integrators and 11 Independent Software Vendors (ISVs). Additionally, the AWS Elemental MediaConnect service team joined the event to provide MediaConnect as one of the cloud gateways used for signal contribution.
Workshop details
The technical theme of this workshop focused on scalability in the cloud with a lens on development operations (DevOps). For DevOps, we focused on two areas. The first was automated infrastructure deployment using Infrastructure as Code (IaC). The second was observability tools to capture and visualize meaningful metrics and logs from cloud infrastructure, applications, and media assets.
The workshop featured two workflow paths that facilitated new integrations between participants:
- Newsrooms included: Cloud ingress and gateway using AWS Elemental MediaConnect, NRCS using Vizrt and AP, MAM, vision mixing and audio mixing using Ross, graphics and automation using Vizrt, intercom using Telos Alliance, and routing and multi-viewers using Grass Valley.
- Sports included: Cloud ingress and gateway using Evertz, vision mixing using Grass Valley, audio mixing using Grass Valley, routing using TVU, replay using Evertz, commentary using Spalk, intercom using Telos Alliance, and multi-viewers using TAG.
Leading up to the workshop, partners created deployment templates using IaC services and tools, such as AWS CloudFormation and Terraform by HashiCorp. They also integrated with HAQM CloudWatch by installing a CloudWatch agent and publishing metrics and logs (including custom ones) to CloudWatch for monitoring and dashboarding. Furthermore, they implemented their applications with AWS Systems Manager for configuration automation and integration with AWS Fault Injection Service.
Connectivity from the Prime Video facility to AWS used AWS Direct Connect to contribute on-premises video feeds as SRT. The format in the cloud was NDI. As partners support a mix of NDI versions 5 and 6, they were able to test interoperability across the two versions. The video format was 1080p at 59.94 frames per second with 4:2:0 8-bit sampling. Participants received access to content with varying data rates (~6Mbps-120Mbps) and 10bit content for additional media asset diversity.
Testing scalability was executed through:
- Increasing signal counts
- Increasing signal bitrates
- Using AWS Fault Injection Service to stress test applications
The following images show some of the metrics and logs that were published from the participants applications to CloudWatch during the workshop.
Image 1: HAQM CloudWatch custom metrics captured from 2 Windows-based EC2 GPUs.
Image 2: NDI Discovery Server pushing logs to HAQM CloudWatch.
Image 3: NDI Discovery Server publishing metrics to HAQM CloudWatch.
As a stretch goal, participants learned about methods to leverage generative AI for observability. For example, they could use HAQM Bedrock to summarize and gain insights into HAQM CloudWatch logs. The steps to achieve this integration were based on guidance provided in Using Generative AI to Gain Insights into CloudWatch Logs.
Conclusion
As AWS continues to listen to the voices of our customers and partners, working backwards from their requirements, future workshops will continue to address broadcast needs while supporting AWS Partners in their journey through learning and collaboration.
Sincere appreciation goes to the participants who took the time and effort to be part of the event: Alpha, AP, Avsono, Diversified, Evertz, Grass Valley, NDI, Ross, Spalk, TAG Video Systems, Telos Alliance, TVU Networks, Vizrt.
Check out the AWS for Media & Entertainment webpage or contact us to know how we can help accelerate your business.
To learn more about our previous interoperability workshops, visit these blog posts: