Delivering TV Everywhere with Cloud Native Solutions
summary
MediaKind migrates a broadcast stack since 2015 from appliances to microservices / containers / Kubernetes & Helm, running on bare metal, OpenStack and public clouds, and solves real problems: multicast to unicast conversion, CNI latency and CPU cost, multi‑NIC control/data separation, and CDN integration. Production uses overprovisioned 1+1 sync redundancy for live encoders, a predictive ML model for per‑channel resource sizing, custom monitoring including a Grafana video thumbnail plugin, and a shift to stateless tiled encoding for 6K/8K multi‑view/VR workflows.
AI-written summary · about
chapters (17)
- 0:00Intro
- 1:38MediaKind Overview
- 2:24What we do? Media & Entertainment
- 4:29Network & Pay TV operators are under pressure
- 7:06The answer: Media moving Cloud Native
- 9:04What is so specific for Media?
- 11:42Live TV Evolution 2018 - 2020
- 13:03Architecture Evolution
- 14:05What about infrastructure?
- 15:44What about networking?
- 18:00Live TV and Container Networking
- 19:49Container Networking: What a journey!
- 26:00Application deployment
- 28:20What about resource requests?
- 30:06Monitoring TV Operations
- 31:51360° augmented viewing experience
- 34:03Autoscaling: More metrics please!
related talks
22mKeynote: Kubernetes: As Seen On TV
34mDisruption and reinvention of media and entertainment
23mHighly Available Live Encoding Using Go and Kubernetes
10mLive events: Moving on-premises broadcast workloads to the cloud
40mCase Study: Kubernetes at Comcast
22mDistributed Video Management at Tesco Scale with Edge Compute and Kubernetes
54mMigrating to Cloud Native with Microservices
36mUsing Kubernetes for Handling Second Screen Experience of European TV Show