Kubernetes: As Seen On TV

Kubernetes is not just a product to be used solely as it comes out of the box, it’s also a platform, a series of building blocks that can be tweaked, tuned, and modified to fit more specific use cases.

At Comcast we are building a geographically distributed system to stream linear video to millions of cable customers around the country, from 1,000 physical locations, within more than 30 regional networks. We have unique requirements and constraints which have benefited incredibly from the versatility of Kubernetes.

Transcript

1. Kubernetes 
 @erikstmartin Erik St. Martin, Systems Architect, Comcast
10. CAP-1000
11. APEX-1000
15. 3Gbps
17. MPEG-DASH
18. IP Video Benefits • Reduced Bandwidth • More Content • More Formats (4k, etc) • More Devices (tablets, phones, etc) • Ubiquitous IP Network
19. CAP-1000
21. 30+ CRANs / Regional DCs 1,000+ Hub Sites 4 National Data Centers 5,000+ Servers / Blades
22. Requirements • High Availability (No SPOF’s) • Stream / Hubsite Affinity • Static IP’s (Source Specific Multicast) • Live Updates To Stream Configuration • Real Time CPU • Deterministic Software Updates • Easy Debugging & Troubleshooting • Monitoring and Alerting • Ops tooling
31. Require: adzone1: primary or adzone1: secondary Prefer: adzone1: primary Node Affinity
32. Require Nodes without a pod labeled: stream: stream1 Prefer Zones without a pod labeled: stream: stream1 Pod Anti-Affinity
34. Tooling • kubectl get streams • kubectl get streams -l lineup=sports • kubectl get streams -l version,version notin (1.0) • kubectl drain <ip>
36. CPU Real-Time Priority • Streamer requires RT priority for some threads • cpu.rt_runtime must be considered for scheduling • Docker creates a new cgroup for every container • cpu.rt_runtime defaults to 0 • Kubernetes has no knowledge of cpu.rt_runtime
37. Source Specific Multicast / Networking • Requires static IP • IP must be pre-assigned • Layer 3 • Bandwidth based scheduling
38. Closing • Not just an orchestration platform • Can be used as a framework for building distributed systems • Tons of great patterns to learn from
39. Thank You! @erikstmartin