| NSF Org: |
CNS Division Of Computer and Network Systems |
| Recipient: |
|
| Initial Amendment Date: | August 22, 2011 |
| Latest Amendment Date: | August 22, 2011 |
| Award Number: | 1053757 |
| Award Instrument: | Standard Grant |
| Program Manager: |
Darleen Fisher
CNS Division Of Computer and Network Systems CSE Directorate for Computer and Information Science and Engineering |
| Start Date: | September 1, 2011 |
| End Date: | December 31, 2016 (Estimated) |
| Total Intended Award Amount: | $600,106.00 |
| Total Awarded Amount to Date: | $600,106.00 |
| Funds Obligated to Date: |
|
| History of Investigator: |
|
| Recipient Sponsored Research Office: |
341 PINE TREE RD ITHACA NY US 14850-2820 (607)255-5014 |
| Sponsor Congressional District: |
|
| Primary Place of Performance: |
341 PINE TREE RD ITHACA NY US 14850-2820 |
| Primary Place of
Performance Congressional District: |
|
| Unique Entity Identifier (UEI): |
|
| Parent UEI: |
|
| NSF Program(s): | Special Projects - CNS |
| Primary Program Source: |
|
| Program Reference Code(s): |
|
| Program Element Code(s): |
|
| Award Agency Code: | 4900 |
| Fund Agency Code: | 4900 |
| Assistance Listing Number(s): | 47.070 |
ABSTRACT
High-bandwidth, semi-private optical lambda networks carry growing volumes of data on behalf of large datacenters, both in cloud computing environments and for scientific, financial, defense, and other enterprises. These networks are treated as if they are perfect. However, it is not uncommon for them to suffer from unexpected performance degradations and loss rates, even when traffic is well-below capacity. This project studies end-to-end characteristics of lambda networks running at high speeds over long distances, with an eye toward better understanding loss, latency variations, and degraded throughput. This understanding is necessary for the design and efficient use of next-generation lambda networks and to enable the transition to cloud computing.
The research described in this project will focus on measuring, characterizing, modeling, and informing new network and protocol design for optical lambda networks. The educational activities will make aspects of network research more accessible to students; promote research initiative among undergraduate and graduate students, and leverage (interdisciplinary) collaborations between Cornell and Howard University, a historically black college and university (HBCU).
Intellectual Merit: Fundamental science requires well controlled experiments and observation. The research described in this project will produce novel network measurement instruments, along with a body of experimental knowledge needed for building networks and protocols that exhibit high performance. The research agenda has three major components.
- Further Development of the Software Defined Network Adapter (SDNA) that the PI has designed. SDNA provides precise and reproducible measurements of an optical lambda network. By achieving extremely high levels of precision, SDNA can shed light on the complexities of flows that traverse high-speed networks.
- Use of SDNA to measure and characterize optical lambda networks. These measurements will help explain observed packet loss at low data rates and reveal new aspects of network behavior.
- Research and deploy new network protocols that enhance end-to-end performance perceived by applications.
New network protocols will allow application developers and datacenter operators to better take advantage of the capacity of the underlying optical networks.
Broader Impacts: The education/outreach agenda is designed to foster an environment for developing a new generation of network researchers who will take a keen interest in hands-on experimentation. There are two major components:
- Research opportunities for undergraduate and graduate students as well as faculty during the summer and in a semester-long class. This will include exploring the design of hardware and software layers for an SDNA apparatus and using SDNA to conduct network research and analyze network measurements.
- A partnership with Howard University (an HBCU) to collaborate on network research and inspire the next generation of underrepresented minority researchers.
PUBLICATIONS PRODUCED AS A RESULT OF THIS RESEARCH
Note:
When clicking on a Digital Object Identifier (DOI) number, you will be taken to an external
site maintained by the publisher. Some full text articles may not yet be available without a
charge during the embargo (administrative interval).
Some links on this page may take you to non-federal websites. Their policies may differ from
this site.
PROJECT OUTCOMES REPORT
Disclaimer
This Project Outcomes Report for the General Public is displayed verbatim as submitted by the Principal Investigator (PI) for this award. Any opinions, findings, and conclusions or recommendations expressed in this Report are those of the PI and do not necessarily reflect the views of the National Science Foundation; NSF has not approved or endorsed its content.
The goals of the CAREER were three-fold: facilitate a better understanding of inter-datacenter networks by designing a high precision network measurement platform; deploy the designed measurement platform at large scale to conduct network measurement experiments; improve the state-of-art network measurement techniques/algorithms with the observed high precision measurement data.
To this end, the CAREER award has been very successful. We completed the first goal via the design and implementation of a software network interface card (SoNIC), which can perform very precise network measurements: http://sonic.cs.cornell.edu . SoNIC has advanced the state of the art in network measurements and has made measurements such jitter, one-way latency, available bandwidth estimation, and even covert timing channels much more precise than previously possible. Results have been published in SIGCOMM, USENIX Network Systems Design and Implementation (NSDI), and SIGCOMM Internet Measurement Conference (IMC).
We accomplished the second and third goals by improving SoNIC’s design and deploying it to the field. In all, we had three generations for the SoNIC platform: The first two generations of the SoNIC platforms used Altera-based FPGA (field programmable gate array) development boards from Hitech Global and Terasic and the third generation of SoNIC used a Xilinx-based FPGA development board, NetFPGA SUME. Since SoNIC can operate on many heterogeneous platforms, we redesigned it in a higher-level language called Bluespec. This redesign enabled us to significantly reduce the time of creating new SoNIC-based capabilities from years to weeks and months. Moreover, to deploy network measurements in the field and at large scale, we have developed a new platform called P4FPGA that allows P4 (programmable protocol-independent packet processors) programs to be compiled to FPGAs in general and to utilize SoNIC in particular. As a result, researchers can write network measurement and network data plane applications in a high-level language and those programs can be compiled directly to hardware (FPGA) and run in the network. See http://p4fpga.org Results have been published in USENIX NSDI, SIGCOMM, and ACM SIGCOMM Symposium of Software-defined network Research (SOSR).
For the third goal, we have deployed SoNIC and made it available within GENI and we have also deployed it within Cornell’s Gates Hall in Ithaca and Cornell’s NYC campuses, Cornell NYCTech and Cornell Weill Medical School.
The NSF CAREER has enabled significant broader impact that especially impacts underrepresented minority (URM) communities. It enabled the creation of the SoNIC Summer Research Workshop whose aim is to increase participation of underrepresented minorities at the Ph.D. level in computer science. More information about the workshop is on the website at http://www.cs.cornell.edu/workshop/sonic/
The ultimate goal for the SoNIC workshop is for all participants to pursue and obtain a Ph.D. We surveyed all past participants in August 2014 and a number of results stood out: 93% of all respondents stated that the SoNIC workshop directly influenced their involvement in computer science and 87% stated that they are currently or intend to pursue a graduate degree (47% Ph.D. and 40% masters). If even a fraction of the participants continue and receive a Ph.D. it would represent a significant impact: In the 2012-2013 academic year, 25 (1.0%) African Americans, 25 (1.7%) Hispanics, and 2 (0.1%) Native Americans received a Ph.D out of 1,442 total Ph.D's earned in computer science in the United States (Computing Research Association, Taulbee Survey, 2015; http://cra.org/wp-contenUuploads/2016/05/2015-Taulbee Survey.pdf ).
We have hosted the SoNIC Workshop for six years during the NSF CAREER and will continue to seek support to host it and impact URM communities.
Last Modified: 03/15/2017
Modified by: Hakim Weatherspoon
Please report errors in award information by writing to: awardsearch@nsf.gov.
