Joint Keynote Address
summary
Claim: a portability layer is infeasible, so Sky runs apps across clouds by matching app requirements to available services via an intercloud broker. Prototypes: Sky Pilot (ML jobs using Ray, Docker/K8s) and SkyPlane (data overlay transfers). Results: fine‑tuning + batch inference cut time 47% and cost 60%, spot‑instance training cuts cost ~70%, a LLM experiment ran for ~$300 on spots. SkyPlane outperforms AWS DataSync. About 20 contributors and several extra clouds joined after early tests.
AI-written summary · about
chapters (17)
- 0:00Introduction to Sky
- 1:28What is Sky Computing
- 3:36Motivations for Sky
- 4:57Challenges of Intercloud
- 8:33The Sky Approach
- 11:15Machine Learning Example
- 14:23Sky Broker Architecture
- 17:07Why Sky Will Happen
- 24:54Sky Development Philosophy
- 28:51Impact of Sky Computing
- 33:27Sky for Single Clouds
- 35:27SkyPilot Prototype
- 39:28Spot Instance Optimization
- 41:48High-Level Service Example
- 45:26Sky Ecosystem Growth
- 46:09SkyPlane Data Transfer
- 49:55Conclusion and Summary
related talks
52mSky Computing
16mSky Computing: Charting a New Course in AI Cloud Infrastructure
17mSkyPilot: An Intercloud Broker for Sky Computing
16mSkyplane: Optimizing Transfer Cost and Throughput Using Cloud-Aware Overlays
11mSkyPilot: Simple Orchestration of Workloads
18mKeynote: Mind the Gap: Bridging Cloud Native Innovation with Real-World Use
18mLucas Welch & Hamish Hill, Skytap | DockerCon 2018
14mKeynote: Science at Light Speed: Cloud Native Infrastructure for Astronomy Workloads