Video Transcoding Reference Architecture

Abstract

This reference architecture provides a concrete example of how to create a scalable, portable, and cost-effective media processing workflow. A traditional Video On-Demand (VOD) workflow is demonstrated in which a source video is output to an online distribution format. The prescribed architecture for this workflow can be cost-effectively scaled and extended to the formats required for your video transcoding use-cases. Some use-cases addressed by this workflow include:

• Supporting content for blogs or social media
• Producing video or digital assets for streaming services
• Embedding content in business applications

Review the video transcoding architecture diagrams for a high-level depiction of a general video transcoding workflow, as well as a more granular version which prescribes specific technologies to implement the workflow.

Technologies Used

The workflow in this document is implemented on the Akamai Connected Cloud (in particular, the Linode Kubernetes Engine), Akamai CDN, and a GitHub Actions-powered CI/CD powered pipeline. The full accounting of technologies used includes:

• Akamai Connected Cloud technologies:

  Technology	Description
  Linode Kubernetes Engine (LKE)	A fully-managed K8s container orchestration engine for deploying and managing containerized applications and workloads
  NodeBalancers	Managed cloud load balancers
  Object Storage	S3-compatible Object Storage, used to manage unstructured data like video files
  Block Storage	Network-attached block file storage volumes
  API	Programmatic access to Linode products and services
  DNS Manager	Domain management, free for Akamai Connected Cloud customers

• Other software and services:

  Technology	Description
  Argo	Kubernetes-native workflow engine
  FFmpeg	Encoding/decoding/transcoding multimedia framework
  PyTranscoder	Python wrapper for FFmpeg
  MediaInfo	Gathers metadata for audio/video files
  GitHub	Git-based managed version control service
  Terraform	Infrastructure-as-code provisioning tool
  Helm	Package manager for Kubernetes
  DockerHub	Container image library
  Let’s Encrypt	Free, automated, open certificate authority
  cert-manager	Cloud native certificate management
  NGINX	Load balancer, web server, and reverse proxy
  Prometheus	Monitoring system and time series database
  Grafana	Observability platform

Business Benefits

• Extensibility: This reference architecture supports a myriad of media output format types and workflow step definitions. It can be configured to output to any device, platform, or audience specification.

• Scalability: This solution supports horizontal scalability by adding more Linodes within the Kubernetes cluster, which enables high throughput. Scaling this solution allows you to process a large amount of content in a short period of time. Scaling can support service launch or marketing campaign requirements.

• Cost-effectiveness: Traditional media workflows have to keep a deployed capacity for peak usage. This reference architecture is built on Kubernetes and uses the Argo workflow engine, which supports dynamic pod scheduling and tear-down. Because of this dynamic resource usage, your cost footprint can be minimized.