Real-time Serverless: Function as a Service with strong QoS

While serverless (FaaS) computing has become wildly popular, all of the major commercial services provide only best-effort service -- no performance guarantees. This makes FaaS unusable for applications requiring any QoS guarantees.
The Real-time Serverless project (RTS) is defining a new class of function-as-a-service (FaaS) that enables applications that require strong quality-of-services (QoS) to enjoy the benefits of the serverless model (stateless design, dynamic scaling, etc.). Real-time Serverless involves the design of new service interfaces that express dynamic performance requirements, such as invocation rates, as well as the myriad system design issues (resource management, admission control, scaling, and even hardware architecture) that are needed for efficient implementation.
Real-time serverless is relevant to bursty, edge applications such as distributed video analytics, virtual reality, augmented reality, and much more. It is also relevant to large scale cloud applications where it can enable performance engineering without resorting to container or VM management. An initial design of the Real-time Serverless design was published in 2019 .
Real-time serverless is part of the Zero-carbon cloud (ZCCloud) project that is pursing a bold, new approach to reduce the carbon footprint of the rapidly growing cloud. It is part of a longer thread of research on volatile resource computing models, such as "spot" instances and preemptible VM's.

- May 2021: HiPS accepts "Motivating High Performance Serverless" paper
- January 2020 News: Real-time Serverless paper wins Best Paper award at Workshop on Serverless Computing!
- September 2019 News: Intel Research funds the real-time serverless part of Zero-Carbon Cloud project!

Publications:
  1. (NEW!) Hai Nguyen, Zhifei Yang, and Andrew A Chien. Motivating High Performance Serverless , High Performance Serverless Workshop at HPDC '21, (Sweden) June 2021.
  2. Hai Nguyen, Kuntai Du, Junchen Jiang, and Andrew A Chien. Dynamic Edge Resource Management for Low Latency Guarantee, in preparation.
  3. (Best Paper Awardee) Hai Nguyen, Chaojie Zhang, Zhujun Xiao, and Andrew A. Chien, Real-time Serverless: Enabling Application Performance Guarantees , Workshop on Serverless Computing (WoSC'19), December 2019, Sacramento, California.
  4. Junchen Jiang, Yuhao Zhou, Ganesh Ananthanarayanan, Yuanchao Shu, and Andrew A. Chien, Networked Cameras are the new Big Data Clusters , Mobicom/HotEdge Video '19, October 2019, Los Cabos, Mexico.
  5. Hai Nguyen, Chaojie Zhang, Zhujun Xiao, and Andrew A. Chien, Real-time Serverless: Cloud Resource Management for Bursty, Real-time Workloads, UChicago Technical Report, March 2019.
  6. Hai Nguyen, Cloud Resource Management for Bursty, Real-time Workloads , MS Paper, UChicago Technical report, June 2019.
  7. Chaojie Zhang, Varun Gupta, and Andrew A. Chien, Information Models: Creating and Preserving Value in Volatile Cloud Resources , in the IEEE International Conference on Cloud Engineering (IC2E), June 2019, Prague, Czechoslovakia.
  8. Chaojie Zhang, Managing the Value of Volatile Cloud Resources: Information Disclosure and Guarantee-Preserving Management , MS Thesis, Technical Report, Fall 2018.
  9. Chaojie Zhang, Varun Gupta, and Andrew A. Chien, "How to Increase the Value of Volatile Cloud Resources: Resource Management and Information Disclosure", Tech Report , April 2018.
  10. Fan Yang, Haryadi Gunawi, and Andrew A. Chien, Resilient Cloud in Dynamic Resource Environments , ACM Symposium on Cloud Computing (SoCC), Santa Clara, Californi, September 2017 (Poster)
  11. Andrew A. Chien, Rich Wolski, and Fan Yang, Dispatchable Computational Loads to Tolerate Renewable Power Generation Variability, Energy Policy Research Conference, September 2015, Denver, Colorado.  In Electricity Journal, October 2015.
People: Hai Nguyen, Chaojie (Sam) Zhang, Kuntai Du, Zhujun Xiao, Junchen Jiang, and Andrew A. Chien , Rich Wolski (UCSB),  Former: Fan Yang, Jeremy Archer (UChicago)