CRII: CNS: Cloud Cartography: Measurement Capabilities for the Modern Internet

Public clouds are ubiquitous in modern life, and their importance necessitates understanding the paths between cloud applications and users to better inform public policy, network operations outside the cloud, and cloud application developers. Cloud wide area networks (WANs) connect users to cloud resources, such as applications, storage, and content, but function as a black box for everyone but the clouds themselves.

This project will enable a comprehensive understanding of the paths between clouds and users outside the cloud, helping to diagnose and locate problems between cloud applications and their users, inform the deployment of applications and content into the clouds, and facilitate comparison of connectivity in the United States to other regions. Furthermore, the ability to accurately interpret paths from users to the cloud offers the potential for developing tools to allow even non-technical users to identify the responsible party for network performance degradations.

The project will also provide undergraduate students the opportunity for hands-on experience with cloud computing and cloud network measurement, through a class designed to engage and mentor women and underrepresented minorities.

This project includes novel techniques to accomplish two separate but related tasks to overcome important shortcomings of current Internet cartography. The first task devises new traceroute probing and interpretation techniques to accurately interpret the network operator for each router on the paths that the public clouds use to reach users around the world, paying special attention to the points of interconnection between cloud WANs and other Internet networks.

The results from Task 1 will provide an unprecedented view of the interaction between public cloud providers and the public Internet, and reveal the paths that cloud applications take to reach their users. Task 2 revisits the decades-old challenge of inferring network interconnections and infrastructure ownership from individual observed paths, a persistent challenge due to fundamental limitations of path measurement capabilities.

The results from Task 1 will help develop, implement, test, and validate a solution to this problem for paths from users toward the cloud.

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.