Cisco Research Center

Request for Proposal (RFP) Response

Review Your Application
Your proposal information appears below. If you need to make changes or are not ready to submit your proposal at this time, click the "Save Only" button at the bottom of this screen. The proposal will then be available to edit from the Welcome page.

Clicking the "Submit" button will immediately send the application to Cisco's University Research Program (URP), and you will then be unable to perform further editing.

NOTE: If we determine that more information is required in order to properly evaluate your request, you will receive an e-mail notifying you that the proposal has been returned to you for revision. Upon receipt of this e-mail, you will be permitted to edit this proposal as requested.
Principal Investigator
Salutation (Ms., Mr., Mrs., Dr., etc.) (required)
Dr. 
First Name (required)
Gurcharan 
Last Name (required)
Khanna 
Job Title (required)
Director of Research Computing 
Your E-mail Address (required)
gurcharan.khanna@rit.edu 
Telephone (required)
585-475-7504 
Telephone Extension
 
Contact Type (required)
Principal Investigator (Research)
 
Proposal Summary
Program/Proposal Title (required)
Advanced Networking Infrastructure Projects for Computing and Collaboration 
Proposal Topic (required)
Other
 
Proposal Topic, if "Other"
video over ip, storage networking, and computer backplanes 
Proposal Abstract (required)
The Research Computing group at RIT is building a research computing infrastructure to support the growing needs of researchers across all disciplines. High performance networks are a crucial part of the overall plan that is targeted to address two specific needs for research use:

* High-speed/low-latency interconnects for high performance message passing and data transfers within clusters and across clusters in grids

* High-speed/low-latency multipoint connections for live sustained uncompressed high definition multicast video streams

Current solutions for these needs include expensive, proprietary systems that are very specialized for their targeted needs. We propose to evaluate commodity, standardized solutions that are less expensive, more universal, and easier to support, such as 10 Gigabit Ethernet. We feel that the performance advancement of such generic networking solutions make them an attractive alternative to traditional dedicated solutions.

We propose to install, benchmark, and evaluate these in terms of a cost/benefit analysis, especially by being able to implement a single solution for these multiple needs. While such standard solutions as 10 Gigabit Ethernet are virtually commodity products now, they are not universally being adopted as readily as one might expect. Our intent is to demonstrate the viability of integrating easily available high performance network components into an advanced research computing infrastructure.

RIT currently does research on advanced collaboration environments (ICE Lab, http://www.rit.edu/~rc/ice.html), grid development (NYSGrid, http://www.nysgrid.org), and networking protocols (NSSA, http://www.nssa.rit.edu). Personnel experienced in these research areas will support and guide the deployment and testing of the proposed advanced networking capabilities. This project is planned to take six months from initial design to first testing of results, with another six months for a complete evaluation. 

Name & Address of Parent Institution
Legal Name (required)
Rochester Institute of Technology 
Address (required)
1 Lomb Memorial Drive  
City (required)
Rochester 
Province
 
State (required)
New York
 
Zip (required)
14623 
Country (required)
United States of America
 
Organization Phone (required)
585-475-7504 
Organization URL
http://www.rit.edu/ 
Primary Investigators, Researchers, & Graduate Students
Lead Principal Investigator (required)
Gurcharan Khanna, Ph.D. 
Lead PI Bio
Gurcharan has a special interest in innovative collaboration tools and the social aspects of technologically connected communities. He started the first Access Grid nodes at RIT and Dartmouth College and he helped start the Internet2 Research Technology Collaboration SIG. He is a member of the Board of the New York State Grid, an advanced collaborative cyberinfrastructure for supporting and enhancing research and education.

Gurcharan is currently Director of Research Computing at Rochester Institute of Technology and reports to the Vice President for Research. His overall function is to encourage and support research at RIT by partnering with researchers to support them with advanced research technology resources. Gurcharan created the Interactive Collaborative Environments Lab housed in the Center for Advancing the Study of Cyberinfrastructure at RIT, as a teaching and learning, research and development, practical application lab, that would have some impact on user interface design and technology choices.

Gurcharan was a Member of the Real Time Communications Advisory Group, Internet2 from 2005-2006. He was formerly Associate Director for Research Computing at Dartmouth College where his focus was on developing and delivering advanced computing technologies to researchers. He directed a six member group from 1995-2004, supporting the evolving IT needs of researchers campus-wide with high performance computers and professional support services. He has served as a consultant on several grant proposals to design and implement multipoint collaborative conferencing systems and twice as a panelist for the NSF Advanced Networking Infrastructure Research Program (2001-2002).

His background includes teaching in the Geography Department at the University of Southern California and teaching and research at the University of California, Berkeley from 1980-1992, where he received his Ph.D. in anthropology. 

Additional PIs
 
Other Researchers Involved
Andy Elble

Infrastructure Engineer, Communication Technology Lead, ITS, RIT

Andy has been in the IT field more than 10 years, all in the higher education arena. Prior to coming to RIT, he was a senior network engineer at Northwestern University in Chicago, Illinois and at the University of Buffalo (NY).

As an infrastructure engineer Andy is involved in what he calls the "technology vision" - breaking barriers around separate systems. Instead of seeing individual systems such as Windows or Linux, he asks, "How do we make this all work as a whole? What is the big picture?"

Some of the current projects he is involved in utilize next generation Internet applications such as Internet 2/Dark Fiber, IPv6, and deployment of the National Lambda Rail at RIT. As RIT grows in its research computing applications and the demand for more web applications, engineers such as Andy are instrumental in building an infrastructure that is both robust and scalable.

-------------------------------------------------

Xiaojun Cao, Ph.D.

Assistant Professor, NSSA, RIT

Dr. Xiaojun Cao received his B.S. from Tsinghua University and M.S. from Chinese Academy of Sciences. In June 2004, he received his Ph.D. degree in Computer Science and Engineering from the State University of New York at Buffalo.

Dr. Cao is a recipient of NSF CAREER Award, 2006-2011.

Research

Dr. Xiaojun Cao is working on modelling, analysis, and protocols/algorithms design of communication networks. Important among these are

Optical Networking, Waveband Switching, Optical Burst Switching

Mobile Ad hoc, Sensor Networks and Security

Optical Wireless Communications

(see http://www.it.rit.edu/~cao for more details.)

 

Additional Researchers and Graduate Students
Student researchers and staff will be hired to carry out the research in the proposal. Students will be co-op workers at RIT, meaning they will be full-time employees of the project for the terms that they are employed, taking time off from their formal studies to devote to this research activity. PhD students in the new College of Computing doctoral program will also be offered opportunities to participate. 
Description of Proposed Research
Proposed Research Description (PDF)
 
Proposed Research Description (Text)
Proposed Projects

1) Compute cluster backplane switching fabric: we are currently in the process of disassembling and moving a 52 node IBM linux cluster to a new location. The current Black Diamond Extreme Blade switches are large, consume lots of power, and run at 1 Gbps. We would like to take this opportunity to replace this switch with a current generation 10 GigE switch and evaluate the performance and behavior under real research computing conditions.

We have every expectation that the improvement over the current technology employed will be significant, but may fall short of other more specialized protocols such as InfiniBand and MyriNet technologies. However, the price/point for a standardized 10 GigE system may prove to be extremely favorable.

2) The Research Data Design Team will be looking at a comprehensive strategy for dealing with research data. This includes data throughput, storage, data management, security, and web access. For the specific area of data throughput, the research computing staff are looking at the performance gain that can be achieved with upgrading to 10 GigE switches and determining what tuning needs to be done to make it optimal. Our goal is put all research systems on 10 GigE networks for transport of very large research data files, both on campus and off-campus via the NYSGrid.

3) MPI (message passing interface) between two different systems across subnets is generally avoided due to high relative latency. This project would evaluate the size of jobs that could be run effectively across subnets by combining two or more clusters in different subnets with 10 GigE switches and routers. The network performance requirements are similar to 2. Successful results would allow much larger jobs to be run by spanning loosely coupled clusters.

4) Real-time uncompressed High Definition video for interactive collaboration. The ICE Lab in CASCI is about to start on Phase II of an advanced collaboration technologies project that is focused on real-time interactive collaboration tools with a special emphasis on high quality video.

We are working with partners at the University of Washington and the University of Southern California to implement and enhance their systems to transport high definition video over IP for real time conferencing. These systems require 1.5 Gbps each way for each participant in the conference for full 1080x1920 video. We will be contributing to the effort to enhance the software and hardware to be more efficient in latency performance as well as user interface design.

This conferencing system will be used for on-campus video connectivity between related research groups as well as for off-campus distance learning and other collaboration projects already in place.

5) Professor Xiaojun Cao in the NSSA Department is currently working on creating improved transport mechanisms beyond TCP and UDP to take advantage of high speed networks over ethernet and IP. His experimental and simulated work will provide actual confirmation and testing of his theories and models on current 10 GigE systems.

Expected Results


We anticipate that a successful evaluation of 10 Gigabit Ethernet implementations in the above projects will result in a less costly, easier to maintain, and more greatly extensible network fabric for computing and collaboration by using advanced CISCO products in place of more highly specialized solutions currently being deployed. 

Optional Images
 
References to Other Research
http://researchchannel.org/tech/ihdtv.asp

http://ultragrid.east.isi.edu/

http://www.gloriad-kr.org/hdtv

https://sitola.fi.muni.cz/igrid

 

Timeframe for Funding and Research Completion (required)
Jul 2007 - Dec 2007 Phase I

Design, hire, setup, and initial test results

Jan 2008 - Jun 2008 Phase II

Implementation, benchmarking, evaluation, and documentation 

Research Cooperation with Cisco
Research Cooperation
Once we complete Phase I of this project, we would like to work with Cisco engineers and developers on optimizing 10 GigE networks for high-bandwidth, long-throughput data flows as well as for low-bandwidth, low-latency message passing computational problems. 
Support Requirements
Request Amount (required)
$78,000.00 
Budgetary Breakdown (required)
Project Manager, 100% FTE, 12 months, $60,000

(or equivalent combination of staff and student Co-ops)

10 GigE PCI-Express network interface cards (Neterion)

@ $3,000 each x 6 = $18,000 

Matching Funds
Uncompressed HD servers (computers, video cards, cameras, displays): anticipated request to RIT for $20,000 internal grant following up on current $86,000 RIT grant for building five collaboration venue nodes on campus

Two full-time student Co-op positions in the ICE Lab to support collaboration infrastructure and tool development ($48,000)

Anticipated upgrade to current IBM cluster with newer cluster and fileserver to take full advantage of 10 GigE interfaces via PCI-Express slots. ($90,000)

 

Equipment Requests (optional)
Equipment Request (Excel)
 
Equipment Request (text)
WS-C6509-E 1 Catalyst 6500 Enhanced 9-slot chassis 15RU, no PS, no Fan Tray B $9,500

WS-SUP720-3B 1 Catalyst 6500/Cisco 7600 Supervisor 720 Fabric MSFC3 PFC3B B $28,000

WS-C6509-E-FAN 1 Catalyst 6509-E Chassis Fan Tray B$495

WS-CAC-6000W 2 Cat6500 6000W AC Power Supply B $5,000

WS-X6708-10G-3C 8 C6K 8 port 10 Gigabit Ethernet module with DFC3C (req. X2) B $37,500

XENPAK-10GB-CX4 64 Cisco 10GBASE-CX4 XENPAK Module C $600 

Misc. Information
Cisco Sponsor (required)
David Jaffe 
Check Payee (required)
Rochester Institute of Technology

(Research Computing) 

Check Address (required)
141 Lomb Memorial Drive

Rochester Institute of Technology

Rochester, NY 14623 

Administrative Contact (required)
Katherine Clark

kacsrs@rit.edu

585-475-7984 

Special Wording Required in Award Letter
 
Planning Team Members
Gurcharan Khanna

Andy Elble

Xiaojun Cao

Paul Mezzanini (Senior System Administrator, Research Computing)

Ed Brannin (Lab Manager, ICE Lab, CASCI)

 

   

Need Support?