CICI: SSC: Horizon: Secure Large-Scale Scientific Cloud Computing

点击翻译按钮获取中文摘要

Over the last decade, public and private clouds emerged as de facto platforms for computationally intensive scientific tasks. Today, huge volumes of many types of scientific data are routinely uploaded to the cloud. A large fraction of this data is privacy and/or security sensitive. Unfortunately, despite numerous advances in network and enterprise security, modern clouds remain inherently insecure. Recent experience shows that well-funded, targeted attacks manage to breach network perimeters of both public and private clouds. Horizon is a novel cloud architecture aimed at providing data and computation security within a scientific cloud. Horizon builds upon three premises: (1) strong isolation on end-hosts, (2) fine-grained isolation in the cloud network, and (3) cloud-wide information flow control. To protect the end-hosts, Horizon develops a new layered hypervisor, and disaggregated virtualization stack with key features of: language safety, software fault isolation, and integrated software verification. To provide secure cloud network environment, Horizon relies on a new network architecture and implements a distributed network firewall, where all network communication and exchange of rights are mediated and controlled by the rules of the object capability system. To protect the cloud data, Horizon develops a set of abstractions and mechanisms to enforce cloud-wide information flow control. In Horizon all data is labeled. The hypervisor mediates all communication of each virtual machine and enforces propagation of labels and security checks for each cloud computation.Horizon aims to provide a practical foundation for developing secure cloud infrastructure suitable for large-scale research workflows that require both speed and security. Horizon will be developed using entirely open-source components, and will be openly available to a broad community of scientists in academia and industry.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.