Table of Contents (powered by https://videoken.com)
0:14:05 A Data-Centric Lens on Cloud Programming and Serverless Computing
0:14:42 Content
0:15:37 Sea Changes in Computing
0:16:04 New Platform + New Language = Innovation
0:17:23 The Big Question
0:19:22 We've been talking about this for a while!
0:19:58 Industry finally woke up: Serverless Computing
0:20:49 Serverless 101: Functions-as-a-Service (Faas)
0:22:01 Serverless & the Three Promises of the Cloud
0:23:11 Three Limitations of Current FaaS (e.g. AWS Lambda)
0:25:15 Still, Serverless Opens the Conversation
0:26:22 A First Step: Embracing State
0:27:39 The Challenge: Consistency
0:28:54 Classical Consistency Mechanisms: Coordination
0:29:31 Coordination Avoidance (a poem)
0:30:34 Why Avoid Coordination?
0:31:55 Towards a Solution
0:34:58 CALM: CONSISTENCY AS LOGICAL MONOTONICITY
0:35:55 We'll need some formal definitions
0:36:41 Intuitively...
0:39:07 Two Canonical Examples
0:42:52 That's interesting. Who cares?
0:46:09 Hydro: A Platform for Programming the Cloud
0:48:27 Anna Serverless KVS
0:50:06 Anna Performance
0:51:57 Autoscaling & Multi-Tier Cost Tradeoffs
0:53:07 Applications on Cloudburst
0:53:54 ModelZoo on Cloudburst
0:54:20 Why Serverless Jupyter?
Large Jupyter deployments!
0:56:19 Beyond operational benefits
0:56:32 Demo: Spiking memory usage
0:57:24 This is only a taste of what's possible.
0:58:07 Scalable Serverless Robot Motion Planning with Cloudburst and Anna
0:58:13 Robot Wants to Get Around Room
0:58:22 Multi-Lambda Motion Planning
0:58:43 Motion Planning Compute Requirements
0:59:13 Motion Planning in Serverless Computing
0:59:46 Lambda Communication
0:59:58 Communication Bottleneck
1:00:09 Overcoming Communication Bottleneck
1:00:34 Cloudburst + Anna: Motion Plan Cost Over Time
1:02:23 Did We Answer the Big Question? Not Yet.
1:03:15 Research Futures
1:03:49 More Information
1:05:06 Q&A