A. Project Summary
The University of Georgia (UGA) in partnership with PeachNet, the State of Georgia higher education data networking service provider, proposes to establish a high performance network link (OC3c level) from the UGA Athens campus to the NSF vBNS Point of Presence (POP) in Austell, Georgia through the facilities of the Georgia GigaPoP Association. The purpose of this link will be to provide regional and national high performance networking connectivity to support a number of worthy research applications on the UGA campus which can benefit from this service. The research activities to be supported include applications in the biological, chemical, physical, environmental and information sciences. In general terms, these applications involve 1) distributed databases 2) remote visualization, instrumentation, and image processing; and 3) telecollaboration, teleconferencing, and distance learning. Each requires qualities of service (QoS) not presently available through the commodity Internet.
The project will leverage investments in ATM based networking facilities currently underway on the UGA campus with similar modernization efforts for PeachNet throughout the State and the Georgia GigaPoP facility located in Atlanta, Georgia. The Georgia GigaPoP has been established to support high performance research and higher education networking connectivity regionally in cooperation with the SURA sponsored Southern Crossroads (SoX) initiative.
Close collaboration between network engineers and applications scientists involved in the project will be required to promote development of the infrastructure to support QoS for assured delivery of video content and real time delivery of data. The entire path from desktop to desktop across the region must support this. The UGA will participate with PeachNet, SoX, and the Internet2 community to construct and support this infrastructure using standard protocols. Since ATM will be available in each of these networks, QoS can be established early on via manual configuration using VBR and CBR PVC's. The UGA plans to work with PeachNet and SoX to implement Internet2 recommended QoS and COS/TOS mechanisms as these become clearly specified. As the vBNS will be delivered through the SoX network, coordination with the SoX engineering team will be a necessary part of this process as well as working with the MCI vBNS engineering team for end to end viability. The close relationship between the MCI and SoX engineering teams is seen as a benefit in using the SoX network to interconnect to the vBNS. in the Southeast.
Overall administration of the project will be vested in University Computing and Networking Services (UCNS) at the UGA.
UCNS is the campus service unit responsible for operation and support of the major shared institutional computing and data
communications facilities at the University of Georgia. Participating research investigator assements will be employed to
evaluate the effectiveness of the services to be provided.
B. Project Description
B.0 Overview of Participating Organizations
p> The partnering organizations submitting this proposal are the University of Georgia (UGA) and the University System of
Georgia Board of Regents Office of Information and Instructional Technology (OIIT). The lead agency in this partnership is
the University of Georgia. Within these parent organizations, the units responsible for performance on the proposal, if
funded, will be University Computing and Networking Services (UCNS) at the UGA and the PeachNet division of OIIT.
UCNS is the campus service unit responsible for operation and support of the major shared institutional computing and data communications facilities at the University of Georgia. In addition to general purpose enterprise servers supporting institutional administrative and academic applications, UCNS operates a variety of specialized computing resources supporting research and student services. These include both distributed memory and shared memory high-performance parallel processing facilities for numerically intensive research computing; a scientific visualization laboratory; an artificial intelligence research and instructional laboratory; specialized computational biology application and database servers; a molecular graphics teaching laboratory, a number of open access microcomputer cluster facilities; and campus email, WWW, and curricular content delivery servers. In addition, UCNS distributes hundreds of microcomputer software products at reduced prices to University departments via site license arrangements with software developers.
Access to campus, national, and international computer networks is provided through the data networking services supported by UCNS. Campus connectivity is provided through the TCP/IP protocol which is supported on server resources operated by the UCNS and a pervasive trunk and building cable system. Within the constraints of available funding, UCNS is undertaking a major modernization of campus trunk networking facilities. This initiative is known as Project Venus (Virtual Electronic Networking for University Services) which is designed to provide the campus with a high speed, optical fiber based network infrastructure that will permit deployment and use of the many advanced networking capabilities called for by the Internet2 project of which the University of Georgia is a charter member. Internet connectivity is provided through PeachNet.
PeachNet is the state-wide, data communications network of the University System of Georgia. PeachNet provides
inter-campus communications and access to external computing resources to each of the University System's thirty-four
institutions, as well as the K-12 and public library communities. Initially deployed 1988, PeachNet's backbone has grown to
be an expansive network serving more than 200 locations connected via high speed (56 Kbps and T1) telephone lines. The network
routes TCP/IP and AppleTalk communication protocols and is connected to the Internet through connections to several national
Internet service providers. A major modernization of PeachNet infrastructure to expand bandwidth capacity and incorporate ATM
QoS capabilities is now underway. This initiative is referred to as the PeachNet2 project. It is driven by the need to support
growing statewide digital library needs, distance education initiatives and telecollaboration activities in a variety of
areas.
B.1 Research and Educational Applications Requiring High Performance Connectivity
Several applications, selected from among approximately 30 Internet2 applications identified on the UGA campus during a
Fall '97 system-wide needs assessment, are seriously limited by existing network capacity; others will be enhanced
significantly through a vBNS connection. These applications are presented as applications worthy of vBNS connectivity in this
proposal. They have been organized into three general categories reflecting their primary network functional requirements.
These are : 1) Distributed Database Applications; 2) Remote Visualization, Instrumentation, and Image Processing; and 3)
Telecollaboration, Teleconferencing, and Distance Learning. A table following the narrative description of these applications
further organizes them according to their vBNS network and QoS requirements.
B.1.1 Distributed Database Applications
B.1.1.1 Fungal Genomics (J. Arnold, Department of Genetics)
In the next millennium, genomics will provide new experimental strategies and technologies to examine global effects of synchronous gene expression on essential cellular processes. Researchers in biology and mathematics will initiate studies with the entire DNA sequence of a model system or pathogen genome; many will need access to this information. The goal of this research is to build an interdisciplinary research infrastructure focusing on fungal genomes. The fungal kingdom includes over 1.5 million different species central to every ecosystem on our planet. They contribute $35 billion annually to the US economy in the form of fermentation products or synthesis of biologically active agents; conversely, they cause an estimated $30 billion in crop loss annually worldwide.
Two fungi in particular will be sequenced in our research. These are N. crassa and P. carinii. University laboratory and computational infrastructures at Georgia, Cincinnati, New Mexico, Texas A & M and Dartmouth are presently used to sequence their genomes and to create functional genomic maps and computational tools for understanding transvection, biological timing, pathogenicity, and chromosomal evolution. Distributed access to and maintenance of genome databases for the scientific community is a major requirement of this research project. Work began at UGA in 1991 on a Fungal Genome Database (FGDB). for constructing and storing physical maps of fungi. The FGDB will receive 10**3 hits per day from a broad range of scientific communities when fully functional. It will accumulate approximately 400,000 base pairs of DNA sequences/day for presentation to/from 13 sites around the world. Anticipated daily transfers will be on the order of 20 GB for updates alone. Additionally, guaranteed packet transport is required for the data to load at mirror sites. If suitable network connections are available a distributed database with concomitant local updating could also be developed. A distributed database system would give us the opportunity to utilize additional computational resources at each participating institution, provide alternate access points to the FGDB, and decentralize curation of the database so that additional resources can be brought to bear for this essential maintenance function. The curators of this database system will be at 8 to 10 sites around the US, including the UGA. They will need access to high-performance tools for dynamically displaying sequence assemblies and physical maps.
B.1.1.2 Astrophysics of Stellar Atmospheres-Project Phoenix (R. Hauschildt, Department of Physics and Astronomy)
Physics and astronomy faculty have developed the fully relativistic, general radiative transfer and stellar atmosphere computer code, PHOENIX, and applied it to a wide range of problems. The code includes very detailed model atoms for a number of important elements as well as several ionization stages for each. In addition, the equation of state includes over 2000 molecules and the capability to simulate dust formation. This means that we can model and accurately simulate the atmospheres of normal stars (ranging from hot stars such as Vega to the Sun and to substellar objects such as Brown Dwarfs and Extrasolar Giant Planets) as well as novae and supernovae. Much of the work of this group involves numerical methods to make these very large calculations possible on modern parallel supercomputers with distributed memory (MIMD architectures). UGA leads the efforts of major collaborators located in Oklahoma, California, Arizona, and Kansas, as well as in France and Germany.
The complexity of the stellar atmosphere simulations is reflected in the computing requirements of PHOENIX. To model the effects of spectral lines on the emitted light of stellar objects, large line lists are required. Currently, our atomic, ionic and molecular line databases contain about 400 million records (spectral lines) with a total file size of about 6GB (using a compressed and encoded data format). In the near future (1 year time frame), we expect to increase this to about 1 billion records or about 15GB. All records of these databases must be processed at least once in each calculation. The databases are also constantly updated when new and better atomic and molecular data become available. The memory and I/O requirements of PHOENIX are also large, typically about 300MB of memory for each CPU of a parallel supercomputer; scratch files of 5GB (an increase to 15GB is projected over the next year) are used in large scale calculations. In order for our research program to progress and to fully leverage our recent advances in computational astrophysics, we need access to reliable high speed network connections. We will need to distribute our databases quickly over the network because most of the simulations will be run off-campus (e.g., at SDSC and NCSA where we have significant allocation of CPU time) and the databases need to be synchronized to central servers (currently located at UGA). Very large calculations will soon exceed the capacity of the systems available at a single location (in fact, this has happened already for some of our experimental models), therefore we will need to use "continental parallel supercomputing" for the upcoming 3D models of stellar atmospheres. This will require database servers to be able to reliably deliver GB's of data very rapidly to distant clients. In addition the clients need to be able to communicate quickly and reliably with each other in order for these simulations to be feasible. For this, we will need guaranteed and sustained high throughput rates for extended periods of time (several hours).
B.1.1.3 GIS Data Clearinghouse Bulk and Graphical Data Distribution Requirements (R. Argo, Office of Outreach in collaboration with Georgia Tech)
The UGA currently houses all of the massive data sets contained in the State of Georgia GIS Data Clearinghouse. A conservative estimate of the size of these data sets is about 300 GB. Several initiatives call for the development of data that will exceed 1.5 terabytes when fully operational. Several Clearinghouse activities require national high performance networking connectivity. Amongst these is serving as a National Geospatial Data Clearing House node to support development and distribution of geographic data based on the Z39.50 search and retrieval protocol to a number of federal agencies and other state planning offices. The Clearinghouse has also entered into a partnership with the U.S. Geological Survey, National Mapping Division, to ulitmately develop core base maps involving transportation, hydrography, wetlands and boundaries for use by this agency. A limiting factor in distributing this data is existing bandwidth. Many of the photo images are compressed for preview only simply because they are too large to view or distribute over existing networks. A major trend in geospatial data delivery is to develop and maintain data in an open GIS format. Key to this development is the ability to exchange data via the Internet. In addition, interoperability of GIS software and data depend heavily on emerging Internet technologies. Maps may include linked photo images and other high volume data that need to be transferred as a unit. Many of the data sets currently held are in the 500-MB to 200 GB range. High bandwidth is especially crucial to this type of application.
B.1.1.4 GARLMER (R. Weigert, School of Marine Sciences)
The Georgia Rivers Land Margin Ecosystem Research (GARLMER) initiative is one of four national collaborative efforts funded by the NSF to determine the transport and transformation patterns of inorganic and organic materials carried from the land into the sea by major coastal rivers. In Georgia, five river systems are being studied. These rivers, all within a 120-mile stretch of coastline, share similar temperature, rainfall and tidal regimes. Each river differs in landscape characteristics, geological setting, flow rate, pH and the amount of inorganic and organic materials entering the river. The GARLMER initiative spans levels of scale from molecular analysis of microbial populations to that of the landscape (employing modern Geographic Information Systems technologies). Massive amounts of data are compiled, collated and ultimately shared among investigators, Coastal Zone Management, Georgia Department of Natural Resources, federal agencies and the public at large through use of the Internet and the World Wide Web. Intensive computer simulation modeling links the research at all levels and scales.
Several aspects of the GARLMER program would be enhanced by connection to the vBNS. Data transfers would benefit greatly. In particular, at the levels of the landscape and above, the geographic information needed involve the transfer of enormous numbers of data files each of which may be many megabytes in size. Their rapid and accurate (uninterrupted) transfer, particularly to the end user from our archives, would greatly increase the usefulness of this type of data presentation. Methods of offering these data graphically and in an interactive manner are available but cumbersome with current transmission speeds. The vBNS connection and Internet II technologies could make these methods into true management decision making tools. All LMER sites are involved with marine systems and collect similar types of data. Efforts are being made by several of the research institutions involved in these initiatives nationally to bring "supercomputing" capability to the desktop by way of linking individual computers over the vBNS. This would greatly enhance the ability to synthesize data sets from all such sites in an effort to find commonalties in ecosystem function across geographic regions. Currently this type of synthesis is tedious at best, if possible at all. Additionally, simulation models to help predict system function and changes due to environmental disturbance could be applied to a broader geographic level by linking the various sites into the same model rather than concentrating on a single locale.
B.1.1.5 AGTEC (L. Pratt, Department of Botany).
The UGA Applied Genetic Technology (AGTEC) Plant Initiative has two overall goals: (1) to bring together diverse expertise and resources in plant biotechnology and (2) to provide facilities that will nurture and stimulate the development and application of these technologies. An essential component of AGTEC is the synergy that either already exists or will be developed among the various research groups within AGTEC, as well as with industrial partners and co-located start-up companies. This synergy will increase the flow of technology transfer from the discovery laboratory to the marketplace.
AGTEC requires advanced networking technology services to support collaborative plant genomics research. Currently, our genomics program includes collaboration among research scientists at The University of Georgia, Clemson University, and the Coastal Plains Experiment Station at Tifton GA. As an example of a current need, the mRNA (gene messages) will be collected from both healthy and diseased plant tissue that has been inoculated with various pathogens or infested with insects or nematodes. These mRNAs are used as templates to create cDNA. Essential to this progress is the ability to transfer large databases containing DNA sequences generated by these genomic programs. It will require both uploading and downloading genome databases from different sites in the USA, Europe, and Asia. As an example, these databases will consist of 400,000 to 1,000,00,000 expressed sequence tags (ESTs). Each EST will consist of 350 to 450 bases. These ESTs will be made available to scientists from around the world for searching, retrieving, and conducting statistical analyses. In many cases, searches will be done in centralized databases, require uploading the sequences and downloading the results. It is important that these databases are transmitted without error. AGTEC scientists will also generate a large number of population-specific genomic maps. This will require use of extensive graphics and high-speed transmission. Current transmission rates severely limit the graphical delivery of existing genomic maps. Researchers at the University of Georgia, Clemson University, and the Coastal Plain Experiment Station also require telecollaboration to coordinate their efforts It is important to bring together on any given day scientists with expertise in molecular biology, pathology, nematology, entomology, and whole plant physiology along with high quality graphical transmission of the plants phenotype in "real time."
B1.1.6 Georgia Structural Biology Initiative (B.C. Wang, Ramsey-GRA Eminent Scholar in Structural Biology, Department of Biochemistry and Molecular Biology).
A major component of the Georgia Structural Biology Resource (GSBR) is the University of Georgia BioCrystallography Laboratory (BCL). The BCL currently operates the first Official Protein Data Bank (PDB) Mirror Site in the US where the entire contents, 8000+ structures of the PDB are made available over the internet to researchers in Georgia, the southeast and the world at large. The Mirror Site (http://BCL10.bmb.uga.edu) began operation in January 1997 as a response to complaints by researchers in the southeast who were experiencing network problems when using the PDB's www server at Brookhaven National Laboratory (BNL), Upton NY. The contents of the PDB are updated weekly with ~100 new protein structures added in each update. The Mirror Site also provides browsing and graphic tools to aid researchers in identifying and visualizing protein structures. In the next millennium, the PDB expects to receive over 200 structures per week and include about 25,000 entries by the year 2004. To meet this increase, the PDB has implemented a new web based deposition and verification procedure and we expect that some of this processing will be off-loaded to the mirror sites as the demand at the BNL site increases. The PDB is already experiencing ~3000 hits per day. The BCL also relies on structural biology WWW servers located around the world for software maintenance, sequence analysis, structure prediction, structure verification and literature searches. Network performance impacts all these areas and improved network performance will lead to greater research efficiency and productivity. Thus, the high bandwidth and low packet loss, which the vBNS will provide is essential to the future operation of the UGA Mirror Site.
University of Georgia has also taken the lead in forming the Southeast Synchrotron Consortium (SESC) whose purpose is to
build and operate a state-of-the-art synchrotron beamline at the Advanced Photon Source (APS), Argonne National Laboratory.
SESC is composed of 59 research groups from eight (Alabama, Florida, Georgia, Kentucky, North and South Carolina, Tennessee
and Virginia) southeastern states. The APS first beamline is scheduled to go online in mid 2001 and will produce data at a
rate of 5-30 MB/second depending on the detector size and exposure time used in the experiment. When the second beamline
becomes available in mid 2002, these rates could easily double. Once the data are collected it must be processed and archived
in order to accommodate the next user group. Thus, fast and reliable network transfers of raw and processed data to remote
sites in the SESC member states would allow for remote archival and processing of data freeing up valuable resources at the
APS for the next users. Again, the high bandwidth, low packet loss and guaranteed packet transport as that offered by the
vBNS would be essential for successful remote archival and processing of the APS data.
B.1.2 Remote Visualization, Instrumentation & Image Processing
B.1.2.1 Remote Astronomical Observation Project (J. Scott, Physics)
UGA is a principal partner in the Southeastern Association for Research in Astronomy (SARA), a consortium of five universities which, among other things, runs a 0.9 m telescope on Kitt Peak in southern Arizona. The telescope is being prepared to run remotely from any of the five sites and as such a secure flow of about 2 KB/sec for 6-12 hours/night will be required for 50 60 nights/year. UGA is the official archive site of all the data to be gathered with the 0.9-m telescope. Thus, users at the other universities will send their data to UGA over the network. The expected data rate will be 0.5 1.0 GB per night for perhaps 200-240 nights a year. In the near future, this observation mode will also be used for satellite observations and for large national observatories (e.g., NOAO, CTIO, Keck).
Remote observing requires secure reliable connections to control the telescope and to download the images. Typically, 2-20GB of data are generated during each night of observing (depending on the telescope). It is very important that the network connection to the telescope is reliable and have a sustained throughput of about 1MB/s. A loss of connection would cause the telescope to enter a 'safe-mode' (e.g., closing the dome, moving into a default position). It would require roughly 60 minutes to restart observation after the telescope entered safe-mode, which would amount to a substantial loss of observing time (sometimes allocated in chunks of 1 hour). Significantly enhanced connectivity is needed both to advance the collaborative efforts among SARA universities and to enable SARA's resources to be shared extensively among the research community.
B.1.2.2 NMR in Structural Biology (J. H. Prestegard, Eminent Scholar in Biochemistry)
High-field nuclear magnetic resonance (NMR) is fast-becoming a primary structural biology research tool. It offers potential applications in a variety of environments where macromolecules function, including the surfaces of biological membranes. It does not require formation of a crystal as a prerequisite to study and can provide information about dynamics as well as structure. However, very high field spectrometers are expensive resources that are best used collaboratively. The State of Georgia (Georgia Research Alliance) has invested in a high field NMR facility (800 MHZ spectrometer) that will serve as a resource for researchers in the state and beyond. This resource, which is located in the Complex Carbohydrate Research Center (CCRC) at UGA, represents an investment in excess of $4M. Sharing these expensive facilities can be expedited through use of high throughput, reliable network connections. While samples could be shipped to the facility, real-time examination of data sets by the originating investigator is essential for efficient use of the facility. Even raw data sets can be large (16-64MB) and transfer rates under current conditions are prohibitive.
Recently, transfer rates have been measured for the same NMR data set between computers within the CCRC, the CCRC and another UGA building, the CCRC and Georgia State, the CCRC and Yale University, and Yale University and Georgia State. The results suggest that UGA-linked transfer rates are currently too slow to allow examination and feedback on intermediate data sets, taking more than 20 minutes in some cases. In contrast, the transfer rate between Yale and Georgia State (both vBNS connected) was nearly 10 times faster than CCRC and Georgia State. Improved connectivity would enhance both intra-state (e.g., Georgia Tech, Georgia State) as well as inter-region (e.g., Yale) collaborations.
B.1.2.3 Internet Access to Digital Video Library Collections (W. Potter, University Librarian)
The UGA Libraries provide electronic access to an increasing number of its collections. So far these collections have included mostly text and static visual materials. Our next phase of public electronic access to collections will require multicast video streaming. Our Media Archives and Peabody Collection form the largest broadcasting archive in the country, with over 86,000 titles. A major collection within the Archives is the WSB Newsfilm Collection. Over 5 million feet (2,030 hours) of newsfilm, dating from 1949 to 1981, represent a visual history of Atlanta and the Southeast during a period of growth and social change. This collection consists of 70,000 news segments ranging in length from 10 seconds to 20 minutes, with the average being 50 seconds. The University of Georgia holds all rights for the distribution and use of these materials. This is a unique historical collection for students, researchers, and media producers.
We will most likely present these clips as either static MPEG-1 files or streamed video. Since the average clip is 50 seconds, the average file size will be a little over 8 MB. The largest clip is 20 minutes long, so that file size is 200 MB. The total collection will take over 1.2 terabytes of storage. In order to provide reasonable access, a minimum of 20 simultaneous users must be supported. Current bandwidth and transmission quality is inadequate for this service, particularly if the material is to be available for research and/or classroom use.
B.1.2.4 Collaborative Immersive and Virtual Reality Learning Environments Project (M. Jacobson, K. Hay, & M. Hannafin, Eminent Scholar of Technology-Enhanced Learning)
Advanced virtual and immersive technologies support qualitatively advanced ways to represent, analyze, and understand both complex scientific information and human knowledge. These technologies can also enhance student learning activities at all academic levels. This project would utilize state-of-the-art virtual and immersive technologies such as those being developed at NASA, University of Houston, and George Mason University as part of the NSF funded ScienceSpace project. This research has documented significant improvements in scientific understanding resulting from learning activities involving virtual reality (VR) ScienceSpace materials (e.g., NewtonWorld, PaulingWorld, MaxwellWorld). Recent research has emphasized the linking of network-distributed VR ScienceSpace worlds that are collaboratively experienced by multiple students in different locations.
High-speed vBNS access would allow real-time research involving linked virtual reality communities between researchers
at UGA's Learning Performance and Support Laboratory and groups such as the NASA, University of Houston, and the George Mason
ScienceSpace research team as well as the virtual reality group at the University of Illinois NCSA. It would enable
inter-institutional research related to the roles of emerging immersive technologies in the learning of complex scientific
knowledge.
B.1.3 Telecollaboration, Teleconferencing, and Distance Learning
B.1.3.1 Real Time, 2-Way Live Interactive Consultations (R. Fayrer-Hosken & J. Moore, Large Animal Medicine)
The ability to interact in real time over distances is essential to the future of clinical and research activities. The second generation Internet's faster transmission speeds, improved transmission qualities, and ability to support 2-way full-screen interaction involving motion and sound will allow faculty members and students at colleges of veterinary medicine around the world to collaborate immediately in the clinical evaluation of animals. In this regard, the theriogenology faculty need to interact with colleagues at Texas A&M's College of Veterinary Medicine regarding cases of equine reproduction and specifically stallion infertility. The clinical evaluation of stallions in Georgia, and digital images of the stallion's sperm cell motility and morphology will be shared with experts at Texas A&M. The result will be enhanced diagnostic evaluations of the animals, improved collaboration among universities, and unique exposures for veterinary students on both campuses.
Our theriogenology faculty also wish to enhance their research on immunocontraception by interacting with personnel at participating zoos. Specifically, the results of on-going immunocontraceptive research will be shared between the College of Veterinary Medicine and participating national zoos. The transrectal ultrasound images of the elephants at the zoos will be evaluated jointly by veterinarians at UGA and other institutions. These capabilities will permit more rapid dissemination of information and allow collaborative evaluation of the reproductive status of the animals involved in the studies. Real time transmission of voice and visual information between veterinarians and faculty in college of veterinary medicine could allow animal owners in distant locations to have their animals examined by renowned faculty members in Athens or other geographically distributed sites. It would facilitate subsequent evaluation of the animal by both local and distributed faculty members and the attending veterinarian.
B.1.3.2 Collaborative TeleConsulting for Healthcare (CaTCH) (A. Sheth, Large Scale Distributed Information Systems Laboratory (LSDIS), Department of Computer Science)
The CaTCH project involves integrating multimedia patient data and medical reference data with desktop video plus data conferencing (for real-time mode) or Web-based streaming media (for asynchronous mode) using WWW/Java programming to establish remote environments permitting context sensitive collaboration management. Our system software has been used to prototype and evaluate a real-world application to support Pediatric Echocardiograph consultation in collaboration with physicians and specialists at the Medical College of Georgia. Additional healthcare applications are planned. Real-time versions of CaTCH need assured high-transmission speeds for effective real-time healthcare consultation applications. The asynchronous versions of CaTCH need specified packet loss rates when acquiring and transmitting video and high-fidelity medical image database records. These are not adequately supported in the existing network, but should be significantly enhanced via vBNS connections.
B.1.3.3 IP Voice-Video Project (J. Segers, Regents Office, University System of Georgia)
Expanded IP-based service, including both voice and video over the Internet, has been the focus of considerable recent national interest. Historically, both voice and video have been carried over time sensitive media. While many experiments over packet base transport systems have been conducted, few installations approach production service levels. The University System of Georgia, as a key collaborator in the Southeast University Research Association (SURA) IP over video initiative, intends to examine the scalability of voice and video over IP throughout a large and diverse university system. This project proposes to take current state-of-the-art software and hardware from major equipment vendors and create a broad scale deployment that will be the foundation for a production system in Georgia with extensions to other institutions nationwide. We will be seeking to collaborate with other institutions around the country that are implementing similar systems using the defined standard protocols. Once identified, we will link the systems in order to examine the scalability of the solution in near production environment.
In the voice portion of the proposal, voice processors will be installed on University System campuses in as many as
fifteen local dialing areas within the State of Georgia starting with UGA. Each processor will be connected to the PeachNet2
Infrastructure and to the local Telephone Company in the area. Each voice processor will be programmed to understand the PTSN
dialing plan; once the system is operational approximately 350 senior level faculty and administrators will be given access
codes. The access codes will permit the customer to use the phone in their office to make a toll free call to anyone one in
any of the other 14 dialing areas. The video portion of the project will proceed on a parallel path. After a multi-site pilot
already underway validates the concept of Video over IP using defined standards, we will extend the effort to 100 other sites
by adding a video over IP converter to each of systems and connecting them to the local PeachNet access point. To preserve
connectivity with the existing system, we will install automatic gateways to the existing system and to the ISDN video
environment.
Meritorious applications and their primary network (shaded) and QoS
requirements.
| Project Title (Research Unit- Project Director(s)) |
Current and/or
Projected Collaborator(s) |
Primary QoS
Priorities |
Distributed Database Applications | Remote Visualization, Instrumentation, Image Processing | Telecollaboration
Telementoring, Distance Learning |
| 1.1.1: Fungal Genomics
(Genetics-Arnold) |
U. of Cincinnati & New Mexico, Texas A&M, Dartmouth U. | 1, 2, 4 | |||
| 1.1.2: Project Phoenix
(Physics-Hauschildt) |
U. of Oklahoma & California, Arizona, Kansas, NCSA, SDSC | 1, 2, 3, 4 | |||
| 1.1.3: GIS Clearinghouse (UGA/ITOS-Argo; Georgia Tech) | State of Georgia, U. System of Ga., federal and other state agencies | 1, 4 |
| 1.1.4: GARLMER (Marine Sciences, R. Wiegert) | Other LMER sites, Dep't of Natural Resources, federal environmental agencies | 1, 2, 3 |
| 1.1.5: AGTEC
(Botany, L. Pratt) |
Clemson U., Coastal Plains Experiment Station (Int'l sites in Europe and Asia) | 1, 3, 4 | |||
| 1.1.6: GSBR (Biochem. & Molecular Bio., B.C. Wang) | SE research universities, Argonne, Brookhaven | 1,2,4 | |||
| 1.2.1: Remote Astronomy
(Physics, J. Scott) |
SARA, NOAA, CTIO, Keck Peak Observ., UCAR | 1, 2, 4 | |||
| 1.2.2: High-field NMR
(Biochemistry & Molecular Biology- Prestgard) |
Yale U., USG and other research universities | 1, 3, 4 | |||
| 1.2.3: Digital Video Library (Library-Potter) | University System of Georgia, SURA Libraries | 1, 2, 4 | |||
| 1.2.4: Immersive
and Virtual Reality (LPSL-Jacobson, Hay, Hannafin) |
NASA, U. of Houston, George Mason U., NCSA | 1, 3, 4 | |||
| 1.3.1: Interactive Consultation
(Vet. Med.-R.Fayrer-Hosken & J. Moore) |
Texas A&M U. | 1.3.4 | |||
| 1.3.2: CaTCH (LSDIS/CS-Sheth | Medical College of Georgia | 1,2 | |||
| 1.3.3: PeachNet II (USG-Segers) | University System of Georgia, SURA/SOX IP Over Video | 1,2,3,4 |
NOTES: Darker shading indicates increased importance of the network need for each project;
1=high bandwidth; 2=low packet loss; 3=low/bounded latency; 4=high throughput
B.2 Contribution to Emerging National High Performance Networking Infrastructure
The University of Georgia was one of the initial Internet2 members and is committed to developing an advanced networking infrastructure for its campus and participating in research and development efforts into QoS and Ipv6. More fundamentally UGA believes that the successful deployment of next generation Internet services in support of research and education must be built on a foundation of cooperative relationships involving a natural hierarchy of state and regional networking suppliers. Amongst the reasons contributing to this belief has been the generally prohibitive cost of deploying high speed data circuits to individual campuses which are not located in or near the metropolitan points of presence for high performance networking service providers. This is the circumstance for the UGA campus. The advantages in pooling financial resources and purchasing influence with our partnering organization PeachNet to overcome this barrier has been recently affirmed. This approach also permits us to leverage our expertise in advanced networking technologies with that of our colleagues to the mutual benefit of all. Moreover, it is only through these forms of collaboration that the transfer of advanced networking technology to the wider educational community committed to by the Internet2 project can be realized.
It is from this perspective that our proposal is deemed to contribute significantly to the emerging national high performance network infrastructure. It provides the management structure and organizational affiliations to bring together the resources of a major southeastern research university with those of one of the nation's most successful state education networks and the recently established Georgia GigaPoP Association to accomplish regional connectivity with other research institutions in Georgia, the southeast and nationally through the NSF vBNS. The operational and support arrangements that this cooperative connectivity environment will require for success should be of significant value to others contemplating similar strategies.
In addition to its Internet2 involvement, the UGA is currently participating in the Southern
Crossroads (SoX) project sponsored by the Southeast Universities Research Association (SURA).
Potential benefits from using the interconnectivity of the SoX network and the vBNS are outlined in
the applications section and include regional collaborations in the physical, biological and
environmental sciences. These projects will require close collaboration between network engineers and
applications scientists to promote development of the infrastructure to support QoS for assured delivery
of video content and real time delivery of data. The entire path from desktop to desktop across the
region must support this. The UGA will participate with PeachNet, SoX, and the Internet2 community
to construct and support this infrastructure using standard protocols. Since ATM will be available in
each of these networks, QoS can be established early on via manual configuration using VBR and CBR
PVC's. The UGA plans to work with Peachnet and SoX to implement Internet2 recommended QoS and
COS/TOS mechanisms as these become clearly specified. As the vBNS will be delivered through the
SoX network, coordination with the SoX engineering team will be a necessary part of this process as
well as working with the MCI vBNS engineering team for end to end viability. The close relationship
between the MCI and SoX engineering teams is seen as a benefit in using the SoX network to
interconnect to the vBNS.
B.3 Network Engineering Planning Process, Participants and Plan
The engineering planning process on which this proposal is based has occurred in three dimensions reflecting the levels of networking infrastructure to be integrated. The first is the design and deployment activity which has occurred on the University of Georgia campus. Second are those activities involving PeachNet and the Georgia GigaPoP Association, and finally the regional planning activities required for the SOX initiative.
High performance network planning at the UGA began approximately three years ago. Initially this planning process involved developing operational and performance specifications on which to build a modern and pervasive ATM matrix network for the campus known as Project Venus. In addition preliminary planning resulted in the acquisition of prototype ATM switching equipment for use in test beds to develop staff expertise in cell switched networking. The network specifications resulted in the selection of IBM Global Services as a network design engineering partner through a competitive solicitation. A campus Project Venus design steering committee was established to coordinate and direct the design activities. This effort resulted in a detailed logical and physical engineering design for a campus fiber optic network infrastructure to serve over 200 buildings. The design ultimately will enable implementation of a one gigabit ATM mesh network. A major component of the design was validating enhanced networking requirements through extensive interviews with campus user groups. Deployment of this network is now underway. Management of the installation activities is vested in University Computing and Networking Services (UCNS). One of the PI's on this proposal, W. McRae, directs this campus service organization. A Project Venus Implementation Committee has been established to advise and assist in the deployment effort on which various campus constituent groups are represented as well as a PeachNet liaison.
To facilitate coordinated high performance network planning at the State level, the UGA, Georgia State University (GSU), the Georgia Institute of Technology (GIT) and the OIIT initiated the Georgia GigaPoP project approximately two years ago. This project has now evolved into the Georgia GigaPoP Association which is being formally chartered as a 501.c(3) organization. A steering committee consisting of the chief technology officers from each of the participating organizations was established. Two planning task forces were created under this steering committee. One was an Engineering/Architecture task force and the other an Application Assessment task force. Senior networking design and management personnel from each of the participating organizations constituted the Engineering/Architecture task force which was chaired by R. Hutchins of GIT. The Application Assessment task force was similarly constituted with senior information technology client support management personnel under the leadership of one of the proposal PI's, M. Hannafin. The Engineering/Architecture task force evolved the architecture which has now been implemented for the Georgia GigaPoP through collaborative planning sessions. In this design process, QoS and connectivity requirements determined by the Applications Assessment task force were incorporated.
The Southern Crossroads (SoX) is a cooperative initiative by the members of the Southeastern University Research Association (SURA). The goal of SoX is to facilitate access to current and future highly integrated, digital communications services for education, research and economic development within the region and across the U.S. Management of the planning activities for SoX is vested in two committees. The SoX Advisory Committee is composed of technology leaders form various SURA members. The SoX Management Team is led by R. Hutchins of GIT and includes network managers from other participating SoX organizations.
The schematic below shows the proposed IP over ATM physical connectivity which is planned. The main switching and routing facilities (shown below the line) are being created in Atlanta as a part of the SoX and PeachNet2 projects. The facilities above the line will be a joint effort between this proposal and the PeachNet2 project already underway. The three aggregation points for the PeachNet GigaPoP in Atlanta are located in buildings that are several miles apart and interconnected with State owned fiber operating in a redundant OC-12 configuration. This diversity of locations with a high-speed interconnect offers redundancy in power and environmental services while simultaneously placing connection points near the primary fiber terminations for both state and privately owned fiber. The most important of the three locations is housed on the GIT campus where the primary aggregation point for both the PeachNet and the SoX project is located. These co-locations serve as the primary interconnection point for research and education networks in the southeast and is the location of Georgia's largest network research facility.
The connections above the line are primarily for the UGA located approximately 60 miles northeast of Atlanta. To reduce the risk associated with a single path we plan to provide a redundant path through the Medical College of Georgia located in Augusta, Georgia, some 90 miles to the southeast.
While the physical connections shown are important, the logical connections between the
routing elements are critical to the success of the project. These logical connections form the basis for
the routing decisions that will create the Quality of Service necessary to the success of the proposed
research. There are presently two methods proposed for this activity, the simplest provides a peering
between the UGA campus router and the PeachNet router located on the UGA campus with OSPF used
to share routes for all traffic. This places the decision for correct packet flow on the GigaPoP router
in Atlanta and reduces the route maintenance at UGA. The other proposal is for peering between the
UGA campus router and the PeachNet router on the UGA campus for Internet 1 traffic with a separate
peering between the UGA campus router and the SoX router for vBNS traffic. This will provide
access to all routes directly to the UGA in the same way these routes are presented to the GIT, thus
placing the burden of routing table maintenance and routing decisions on the staff both at UGA and
GIT. In both cases the Quality of Service requirements will be met by accessing the QoS parameters
of the underlying ATM infrastructure.
B.4 The University of Georgia Network Infrastructure
The UGA networking infrastructure described herein is the Project Venus ATM mesh facility now being deployed to supplement and ultimately replace an existing and pervasive Ethernet/MAP campus network. It is through this network environment that connectivity for the worthy applications cited in this proposal will be supported.
The Venus infrastructure is based on a 16 switch node matrix topology incorporating growth
capacity for four additional switch nodes. The initial configuration implements dual 155 Mbps
connections between switch nodes with migration to 622 Mbps permitted as traffic warrants. Four of
the nodes are primary nodes as a function of required connectivity and switching capacity. These four
nodes constitute the ATM backbone network. Each of the remaining secondary nodal locations are
connected to at least one of the primary switch nodes. A cluster of approximately 12 buildings is
associated with each nodal location, both primary and secondary. Each building will be connected
individually and independently in a star topology to their associate cluster node. The switch nodes
provide transport for mixed data, MPEG2 video, and voice traffic. Legacy traffic from client/server
applications using IP, IPX and AppleTalk protocols in the buildings over 10Base-T or 100BaseT
Ethernet to local multiprotocol hubs is supported. Bridges or switches convert frames to cells for
transport over the ATM backbone. When ATM native applications are developed, they will
communicate directly over the ATM network with a Quality of Service (QoS), while the non-ATM
applications will use the ATM Adaptation Layers 1, 2, and 5. The switch nodes will provide QoS for
classes CBR, VBR-RT, VBR-NRT, UBR, and ABR with priority queuing for QoS delivery.
Connectivity to PeachNet services and the OC3c channel for which support is solicited in this proposal
will be provided through the primary switch node located in the secure central computing and
networking facility. Appealing features of this design are:
In anticipation of Project Venus activation, all replacement or new building cable plants installed during the past three years on the UGA campus have complied with institutional structured wiring standards using fiber optic cabling for building backbones and Category 5 unshielded twisted pair (or Cat 5 UTP) cabling to connect end devices. Building hubs acquisitions have also reflected this development.
The following schematic illustrates generically the connectivity of legacy LAN or native ATM
workstation in a given cluster building through a nodal switch (in this case a primary node) to the
PeachNet router/switch equipment that will interface to the OC3c link to be installed.
B.5 QoS Requirements and Implementation Methods
Carrying both production best effort traffic and priority network traffic (such as video content)
on the same network is currently a challenge. The UGA plans to utilize the ATM infrastructure
provided locally on campus, through the PeachNet infrastructure, and across the Southern Crossroads
to the vBNS to implement better than best effort guarantees for worthy applications traffic. Early on,
this service will be provided via CBR and VBR permanent virtual circuits manually configured through
the networks. In the future, it is envisioned that multiple mechanisms may be available to provide this
service: switched virtual circuits through the ATM network signaled by RSVP; or the coming
DIFFSERV architecture utilizing the Bandwidth Broker (BB) system to provide assured connectivity or
relatively better than best effort services. As these new architectures are defined SoX plans to
implement these and extend the services into the state and local networks connected. The UGA will be
a participant in resolving technical issues, management issues, and generating applications research
feedback into the system for the furtherance of these projects.
B.6 Technical Expertise in Computer Networking
The University of Georgia and its PeachNet partner in this proposal have extensive experience in providing TCP/IP internetworking services, including coordination with network service providers. The UGA staff and management who will be responsible for local support of the services described in this proposal currently operate a pervasive campus intranet connecting over 13 thousand nodes involving multiple domains, servers and routers. The University also houses and supports major PeachNet router and server equipment. University networking staff have deployed and operated several ATM test networks. One of these networks has now been implemented in a production status. University of Georgia networking staff are active participants in the ATM Forum, ACUTA and Internet2 (UCAID) activities.
Although not generally known, PeachNet is one of the most successful and extensively used education networks in America. This success is due to the vision and management skill of its developers who are also participants in this proposal. PeachNet development began in 1987 to facilitate the delivery of educational services to students in Georgia by providing TCP/IP data communications services for the 34 colleges and universities comprising the University System of Georgia. Today, PeachNet provides Internet access for education across the state from approximately 200 locations including support of research applications, library access and distance education.
The experience and competencies of UGA and PeachNet networking staff involved in this proposal will be augmented by the considerable expertise of our colleagues at Georgia Tech and Georgia State who are our partners in the Georgia GigaPoP Association.
Biographical sketches for the senior designers and management personnel supporting this
proposal are provided in Section E.
B.7 Service Expansion and Future Support Plans
Expansion of the high performance networking services described in this proposal to the University of Georgia campus broadly is a primary feature of the implementation strategy established in the Project Venus design. Expansion of connectivity within the State of Georgia and regionally will be achieved through the modernization efforts now underway for PeachNet and the connectivity strategies pursued by the SoX consortium.
The campus expansion strategy is based upon the phased implementation explicit in the Project Venus design. These natural stages of implementation are to first deploy the fiber and electronics to activate the four primary switch nodes, thereby establishing an ATM campus backbone. This is followed by installation of the fiber to connect all secondary nodes with selective deployment of secondary node switch equipment and cluster building connectivity as requirements dictate and funding permits. Fiber and electronics to connect the four primary nodes is now being implemented with the fiber to provide connectivity for the remaining nodal buildings to be deployed by calendar year end. Installation and activation of switch electronics in the secondary and tertiary nodal buildings will begin next calendar year with projected completion date 18 months subsequently. In this time period those buildings housing worthy research applications will be activated preferentially.
The existing hierarchal network support structure on the UGA campus which consists of
central NIC, NOC, engineering design, cable installation and LAN consulting services complimented
by distributed domain network liaisons throughout the campus will provide continuing support for this
network modernization initiative.
B.8 Cost Effectiveness and Cost Sharing
The cost effectiveness of the high performance networking strategy presented in this proposal is considered to be optimal under prevailing regulatory and communications equipment pricing environments. Several factors lead us to this conclusion.
The quoted cost for the OC3c Athens service for which NSF funding support is being requested was determined only after a difficult and lengthy competitive bidding process administered by the State of Georgia. The lowest resulting bid for the service was $8 thousand per month. This same bidder had only 18 months previously offered to provide an Athens to Atlanta DS3 link for $28 thousand per month. The other bids received for the OC3c service ranged from approximately $18 thousand to $110 thousand per month. In addition, the switch and hub electronics costs for the UGA campus have been minimized through preferential vendor discounts resulting from competitive bidding inducements.
The proposed strategy also leverages significant investments in modern networking
infrastructure and human resources already undertaken by the UGA on its campus, PeachNet within
the State of Georgia and the Georgia GigaPoP Association regionally in the design and deployment of
ATM networking facilities. On the University of Georgia campus network modernization investments
are conservatively estimated at $2.5 million during the past two fiscal years and are expected to
continue at a level of approximately $1.2 million annually for the next three years. PeachNet
investments to deploy a high performance ATM fabric serving all of higher education in the State of
Georgia are projected to exceed $12 million during the current year. The Georgia GigaPoP Association
has successfully implemented an operational regional aggregation point for traffic onto the vBNS used
by at least 13 southeastern research universities. In this process, substantial contributions of vendor
equipment and connection services have been obtained. This grant request will extend two other grant
awards for High Performance Connections in Georgia: the Georgia State award and the Georgia Tech
award which specified supporting other university awards in the future. Finally, the cost effectiveness
of our request to the NSF must be evaluated against the approximate 50 percent real cost sharing
contribution which is being offered by the University of Georgia Research Foundation.
B.9 Evaluation and Dissemination of Results
Three general strategies will be followed in evaluating the effectiveness of enhanced network connectivity for the cited worthy applications. First, and most significantly, a set of specific enhanced network performance goals will be established for each application that reflect the requirements of the investigators and the capabilities of the extant networking technology. Articulation of these goals will be managed by the Principal Investigators with the assistance of Project Venus Implementation Committee members.. The second component of our evaluation strategy will be to utilize available network metering tools to monitor satisfaction of the QoS functionality requirements established for each application. These measurements will be collaboratively undertaken by NOC staff affiliated with the University of Georgia, PeachNet and the Georgia GigaPoP. Finally, assessments from the researchers involved in utilizing or exercising the identified applications with respect to the effectiveness of the services delivered will be solicited on a regular basis. The information gained from these user assessments and the NOC mediated measurements will be utilized by the PI's and the appropriate network management staff as a basis for evaluating the performance of the vBNS connection to be established.
A variety of methods will be used to disseminate the results of the proposed innovative networking applications. These will include descriptions of outcomes and research benefit in discipline specific publications and presentations by the various functional area investigators. It will be our intent to also present and demonstrate successful applications in SURA and Internet2 sponsored forums as well as other conferences which emphasize the benefit of advanced networking technologies to research and education. The University of Georgia will maintain WWW reference material specific to this project in which progress and results are reported. We will publicize the initiative in institutional press releases, progress reports and recruitment materials provided to prospective research faculty and students. Required progress reports will be provided to the funding agency as well.