SHF: Small: Towards High Performance Serverless Edge Computing for Data-intensive Applications

Suny At Stony Brook, Stony Brook NY

Investigators

Abstract

Edge computing has become an indispensable booster for the rapid development of Internet-of-Things. It provides users not only low-delay and high-bandwidth services but also unprecedented security and privacy guarantee. However, the unique challenges of network edge, such as resource limitation and device heterogeneity, greatly increase the difficulty of developing and managing edge applications. Researchers have turned to emerging serverless computing paradigm for efficient and cost-effective solutions. The light-weight virtualization of serverless edge computing largely mitigates the resource shortage of edge environments while providing extra flexibility. Furthermore, edge application developers are completely liberated from complicated management of edge resources, facilitating the portability and short time-to-market of edge applications. Nevertheless, serverless edge computing is still impeded by the inherent natures of network edge. The computation/data geo-dispersion and device heterogeneity in edge environments may introduce long transmission and computation delay to the initialization and execution of data-intensive edge functions. The resource limitation and volatile network conditions further exacerbate the negative impact and increase the risk of network congestion. This project aims to systematically investigate the challenging issues towards building a high performance serverless edge computing framework for data-intensive applications. This project provides a hardware/software co-design to construct a novel serverless edge computing framework for data-intensive applications. More specifically, it consists of following inter-related research tasks: (1) develop a hardware-enhanced architecture for accelerating data processing, transmission and distribution of serverless edge computing; (2) design a data-aware computation orchestration system that adapts to unique edge conditions and the data acceleration architecture; (3) develop a computation-aware data orchestration system that adjusts the data acceleration architecture dynamically; (4) conduct a comprehensive performance evaluation through extensive simulations and implementation of a prototype of the framework. The outcome of this project will not only enhance the performance of many current data-intensive edge functions, but also facilitate the edge adoption for numerous future applications that rely on efficient data processing and transmission. As edge computing is penetrating into all aspects of human lives, this research will have a profound impact on the society. The project trains graduate students and promotes the participation of female students in computer engineering. The important findings of this project will be integrated into a graduate course and disseminated to research community and industry by way of conferences, journals and website access. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.