Griffin: A Storage Service for Granular Edge Applications

The emerging edge computing paradigm is enabling a series of real-time, location-aware applications. The data produced by applications like autonomous driving, collaborative machine learning, and real-time video processing is often only usable in a limited time- A nd space-window. Data placement, sharing, and computation need to happen with low latency and location awareness, using a heterogeneous infrastructure that is not on par with datacenter hardware. In this work we plan to design and implement Griffin, a distributed edge storage service that is aware of mobility, latency and consistency requirements of applications, and that can support storage function programmability for low-latency responses