Phoebus: An Integrated System for End-to-end High Performance Networking

Project Details

Project Lead
Ezra Kissel 
Project Manager
Martin Swany 
Project Members
Ezra Kissel, Martin Swany, Matthew Jaffee  
Supporting Experts
Gregor von Laszewski  
Institution
Indiana University, ISE  
Discipline
Computer Science (401) 
Subdiscipline
11.07 Computer Science 

Abstract

Phoebus is a protocol and service implementation for improving end-to-end transfers to multi-gigabit speeds. Phoebus provides a network “inlay” that augments the network topology with an additional layer of protocol and functionality. Phoebus leverages the ubiquitous TCP/IPv4 implementation on the end hosts, while providing an adaptor to translate to a variety of alternate protocols over the wide area, including creation of virtual networks on the fly. Many data-intensive applications can use Phoebus without modification and without making any system-level changes. We intend to use the FutureGrid resources to further research on alternate protocols and transport technologies such as RDMA on the core of the network to improve network throughput for research and education users.

Intellectual Merit

The intellectual merit of this research is in providing a vehicle to utilize the network as a data movement service. Our system applies and extends the state of the art in network data movement. This approach allows us to realize proven network optimization techniques in protocols and algorithms at scale. In addition, this work builds a bridge to enable future network protocol innovation from the networking community.

Broader Impacts

The broader impact of this effort is in the effect on data intensive science. This system will address the crisis in data movement and will enable unprecedented ease of network data transfer. By lowering the bar for high-performance data transfer, we will broaden the number of scientists who can take advantage of advanced network infrastructure.

Scale of Use

The experiments should require few nodes (2-3 per site) with high performance networking (including RDMA capabilities) and that span across the WAN. Experiments should not take longer than a couple of hours each time.