What is the Future of Serverless Computing?
summary
Claims that serverless (with Knative on Kubernetes) cuts cost and energy use are backed by specifics: cloud grows ~15% and already consumes ~1–1.5% of global energy, so overcommit/defragmentation can save providers and users money. Concrete hurdles: cold start latency is tens of milliseconds today but targets are <10 ms; solutions include freezing containers, faster probing, and runtime projects like Security Guard plus OpenCost and Kubernetes sustainability efforts. Emerging use cases include data/AI (KFServing/KF) and quantum serverless workflows.
AI-written summary · about
chapters (9)
- 0:00Start
- 6:01motivation
- 7:02historical context
- 8:56managing costs
- 14:14quantum serverless
- 15:38design
- 16:38solution (specifics)
- 17:30threats to validity
- 18:50summary
related talks
40mWho Needs Serverless?
10mEnergy-Efficient and Cost-Optimized Serverless Computing
35mWhat does Serverless mean for tomorrow’s abstracted infrastructure?
41mIs Serverless the End of Kubernetes?
33mGoing FaaSter: Cost-Performance Optimizations of Serverless on Kubernetes
32mThe Path to a Serverless-Native Era with Kubernetes
21mKeynote: Kubernetes Serverless Present and Serverless Future
28mThe Future of Serverless
