A major problem on the Internet is
the scalable dissemination of information. This problem is
particularly acute exactly at the time when the scalability of data
delivery is most important, e.g. election results on the night of the
2000 United States presidential election, and news during 9/11/2001.
The current unicast pull framework simply does not scale up to these
types of workloads. One proposed solution to this scalability problem
is to use multicast communication. However, allowing multicast
communication introduces many non-trivial data management problems,
such as caching, consistency, and scheduling. We aims to build a
middleware that unifies and extends state-of-the-art data management
methods and algorithms into one software distribution. Its
flexible and extensible architecture is built from individual
components that can be selected or replaced depending on the
underlying multicast transport or on the application needs.
Particular care has gone into the design of the algorithms to
optimize the user-perceived level of service.
This project incorporates both applied and
fundamental research investigations. On the applied research side, it
focuses on the design and implementation of middleware for multicast
data dissemination that transparently provides applications with a
flexible array of data management services. On the fundamental
research side, our project seeks to develop optimal algorithms for
each component, and provide an insight into the functional synergy
between the different components. For a more detail description of
our project, see our white paper ( html
, pdf
) .
People
Related Publications
Application
Related Projects
Test Bed
Related Links
Professor Kirk Pruhs , Computer Science Department , University of Pittsburgh
Professor Panos Chrysanthis , Computer Science Department , University of Pittsburgh
Professor Vincenzo Liberatore , Electrical Engineering and Computer Science Department , Case Western Reserve University (CWRU)
Graduate Students at CWRU : Wei Li and Wenhui Zhang, Electrical Engineering and Computer Science Department
Graduate Students at Pittsburgh : Vince Penkrot, Siddartha Roychowdhury, Computer Science Department
Charpters in book: Middleware for Communications , Wiley, 2004.
Application-Perceived Multicast Push Performance , IPDPS 2004.
An Optimized Multicast-based Data Dissemination Middleware: A Demonstration, ICDE 2003.
Middleware Support for Multicast-based Data Dissemination: A Working Reality. (Preliminary version), WORDS 2003.
A comparison of multicast pull models, ESA 2002.
Multicast pull scheduling: when fairness is fine, SODA 2002.
Broadcast Scheduling for Set Requests. In DIMACS Workshop on Resource Management and Scheduling in Next Generation Networks, 2001. Talk slides.
Pushing Politely: Improving Web Responsiveness One Packet at a Time . Tech. Rep. DCS-TR-415, 2000. Abstract. Revised version to appear in PAWS 00, Talk slides.
Exploiting Versions for Handling Updates in Broadcast Disks. Proc. of the 25th Int'l Conference on Very Large Data Bases, pp. 114-125, Edinburgh, Scotland, Sept. 1999.
Scalable Processing of Read-Only Transactions in Broadcast Push, Proc. of the 19th IEEE Int'l Conference on Distributed Computing Systems, Austin, TX, June 1999.
Achieving Consistency in Mobile Databases through Localization in PRO-MOTION. Proc. of the 2nd DEXA Int'l Workshop on Mobility in Databases and Distributed Systems, pp. 82-89, Florence, Italy, Sept. 1999.
Scheduling Jobs Before Shut-Down. Tech. Rep. 99-72, 1999, UMIACS. Abstract. Revised versions in SWAT 2000 and © Nordic Journal of Computing
Data Dissemination on the Web: Speculative and Unobtrusive . Tech. Rep. 99-23, 1999, UMIACS. Abstract.
Caching and Scheduling for Broadcast Disk Systems. Tech. Rep. 98-71, 1998, UMIACS. Abstract. (Revised version to appear in ALENEX 2000). Talk slides [Microsoft Explorer, PNG].
Broadcast Disk Paging with a Small Cache, Tech. Rep. TR98-36, 1998, DIMACS. Abstract.
On Broadcast Disk Paging © SIAM (with Sanjeev Khanna), SIAM Journal on Computing.
Real-Time Outbreak and Disease Surveillance (RODS) is a healthcare alert system developed by the Center for Biomedical Informatics at the University of Pittsburgh.RODS can use our multicast middleware to support the collection and monitoring of the large volume of data needed for the assessment of disease outbreaks, as well as the dissemination of critical information to a large number of health officials when outbreaks of diseases are detected. The demo system utilizes the middleware and is a three tier application. The schema for the database was adapted from the Real Time Outbreak of Disease Laboratory. The database, implemented on one of the department's Oracle servers, was populated with randomly distributed data that would mimic the type of data that is collected in the real system. In particular, cases of disease and prodromes populate the database. For the purpose of this demo, the data was restricted to each of the sixty-seven counties in Pennsylvania. The server application was written in Java 2 using the JDBC to Oracle conduits. The server application queries the database for the counts of each distinct disease and prodrome grouped by county. The server then multicasts these counts in an XML message to the clients listening on a given multicast channel. The client application was written in Java 2 as well and accepts user input via a graphical user interface that displays the map of Pennsylvania broken down by county. When a user clicks on a county, the counts of disease and prodromes in that county are displayed on a separate window in a new table.
CAISR lab at CWRU
Last Modified on Jul. 11, 2004