Preview: Cloud Computing
summary
A set of systems and findings target real limits: DART replaces a centralized master with a distributed hash table overlay to eliminate queuing and long‑tail task latency at the edge; ADFS enforces strict memory caps by almost‑dfs switching between breadth‑ and depth‑first graph execution to trade memory for throughput; CrystalPerf trains ML models from code traces, documentation and past profiles to predict dataflow performance; VM aging links slowdowns to extended page‑table violations; FastNet/FastLane/Sonic use non‑overlapping tree overlays, threading and communication‑aware placement to cut serverless provisioning and interaction latency.
AI-written summary · about
chapters (12)
- 0:00Introduction
- 0:35Overview
- 1:15Stack
- 5:19Runtime
- 6:27Resources
- 7:05Summary
- 7:43Dart
- 8:36Memory Efficient Platforms
- 10:53Performance Debugging
- 12:07Warden
- 12:38Parasync
- 13:39Serverless
related talks
41mCloud Computing: The Edge talks - long road into the Cloud
30mThinner Clouds with Preallocation
15mState of the Infrastructure
18mKeynote: Mind the Gap: Bridging Cloud Native Innovation with Real-World Use
27mFrom Supercomputing to Serving: A Case Study Delivering Cloud Native Foundation Models
17mCloud Computing Made Better: An AWS View on Serverless Computing
56mKeynote on Building the Cloud for You
06mPunching the Clouds