Disaggregated Cloud

点击翻译按钮获取中文摘要

The decades-old approach for building cloud platforms using monolithic servers is inefficient. Monolithic servers pre-package a fixed amount of compute, memory, and storage recourses in rigid servers, limiting the ability to upgrade the resources, tolerate device failures, and efficiently utilize all the resources. The proposed research will explore a new paradigm for designing cloud infrastructure and applications: a disaggregated cloud. The project will explore techniques for building a completely disaggregated cloud infrastructure that first disaggregates cloud hardware resources into network-connected compute, memory, and storage devices and, second, disaggregates the cloud application into a set of server-less lambda functions that can be individually scheduled, allocated, and executed. This project has the potential to widely impact technology, as it explores uncharted territory for designing systems. This innovative project will potentially have a wide influence on technology, including influencing many Canadian companies that build cloud systems (e.g., Huawei, NetApp, Ericsson, and IBM) and the more than 150 Canadian data centers that use these systems. Finally, this program will provide an invaluable opportunity to train highly qualified personnel in cutting-edge networking and storage technology.