UCSD burst-parallel research project

Overview

The general availability by cloud providers of lightweight virtualized containers priced using a fine-grained cost structure provides an opportunity for a new burst parallel computing model.

Our vision is that this new model will enable applications to burst processing into the cloud across hundreds or thousands of concurrent ephemeral “lambdas”, reducing the run time of data processing tasks by multiple orders of magnitude and transforming batch tasks into interactive ones.

Since end users only pay for the time they are using those lambdas, developers can harness this capability without needing many users to amortize an expensive hardware platform. Although originally intended for asynchronous Web microservices, we believe that with sufficient systems support, the same lambda infrastructure, eventually with modifications we propose, is capable of broad and exciting new applications.

People

Group photo (From left to right) Adrian Mendoza (Undergraduate researcher), Zheng Tang (Undergraduate researcher), Elizabeth Guerrero (Undergraduate researcher), Elizabeth Farkas (Undergraduate researcher), George Porter (Associate Professor), Liz Izhikevich (M.S. Student), Shelby Thomas (Ph.D. student), Lixiang Ao (Ph.D. student), Geoff Voelker (Professor)

Publications

Sprocket: A Serverless Video Processing Framework for the Cloud, Lixiang Ao, Liz Izhikevich, Geoffrey M. Voelker, and George Porter, Proceedings of the ninth ACM Symposium on Cloud Computing (Socc’18), Carlsbad, Calif, October 2018.

Dark Packets and the end of Network Scaling, Shelby Thomas, Rob McGuinness, Geoffrey M. Voelker, and George Porter, Proceedings of the ACM/IEEE Symposium on Architectures for Networking and Communications Systems (ANCS), Ithaca, New York, July 2018.

CacheCloud: Towards Speed-of-Light Datacenter Communication, Shelby Thomas, Geoffrey M. Voelker, and George Porter, Proceedings of the 10th USENIX Workshop on Hot Topics in Cloud Computing (HotCloud-X), Boston, MA, July 2018.

Encoding, Fast and Slow: Low-Latency Video Processing Using Thousands of Tiny Threads, Sadjad Fouladi, Riad S. Wahby, Brennan Shacklett, Karthikeyan Vasuki Balasubramaniam, William Zeng, Rahul Bhalerao, Anirudh Sivaraman, George Porter, and Keith Winstein, Proceedings of the 14th ACM/USENIX Symposium on Networked Systems Design and Implementation (NSDI), Boston, MA, March 2017.

Blog