Data Intensive Computing and Communications for Grid and Cloud Computing
Grid and Cloud Scheduling: For many applications, data is distributed across geographically distributed storage systems and owned by multiple organizations. Grid computing allows multiple organizations to share resources. It requires concurrent utilization of a large number of heterogeneous and dynamic resources by multiple applications. With support from a NSF ITR grant and Intel Corporation, I have developed SPHINX, a middleware for managing grid resources that uses novel linear programming based algorithms for resource allocation. These algorithms guarantee application quality of service requirements and organizational policies simultaneously. The SPHINX middleware, was successfully demonstrated at Supercomputing 2003 and 2004.
Control Plane Provisioning for Grid and Could Computing: Communication is crucial portion of scaling data mining and large scale applications on grids. Large scale grids are deploying optical networks to support the high bandwidth requirements of these applications. Realizing these high bandwidths requires careful control of the underlying resources and a good understanding of the underlying topologies. In joint work with another NSF supported project (Ultralight), we are currently developing and analyzing control plane scheduling algorithms for maximum concurrent file transfer problem (MCFTP) in optical networks. We have several linear programming based approaches to solve this problem – a novel aspect of this work is in incorporating spatial and temporal policy constraints. Our preliminary results show that our algorithms can schedule for 100 node networks (with 3-7 edges per node) in several minutes or less.
Fault Tolerant Aggregation in Large Scale Systems: Communication is crucial portion of scaling data mining Our interest is in designing distributed algorithms for aggregation in sensor networks that are reliable, fault tolerant and have good, scalable performance. We have performed in-depth analysis of the techniques proposed for improving the fault tolerance of tree aggregation, such as multiple trees and local fixes. We have also worked on a methodology that optimizes the performance of a hybrid of tree and gossip-based aggregation.
- Yan Li, Sanjay Ranka, Sartaj Sahni: In-advance path reservation for file transfers in e-science applications. The Journal of Supercomputing, Vol. 59(3): 1167-1187, 2012.
- Eun-Sung Jung, Sanjay Ranka, Sartaj Sahni: Topology Aggregation for e-Science Networks. International Journal of Next-General Computing, volume 1, 2010, pp 1- .
- Eun-Sung Jung, Sanjay Ranka and Sartaj Sahni, Topology Aggregation for e-Science Networks, Proceeding of the 10th IEEE/ACM International Symposium on Cluster, Cloud and Grid Computing, pp. 530-533.
- Eun-Sung Jung, Sanjay Ranka and Sartaj Sahni, Bandwidth Allocation for Iterative Data-dependent e-Science Applications, Proceedings of the 10th IEEE/ACM International Symposium on Cluster, Cloud and Grid Computing, pp. 233-242.
- Yan Li, Sanjay Ranka, Sartaj Sahni: First-Slot scheduling with wavelength conversion for distributed file transfers. ISSPIT 2010: 42-47.
- Laukik Chitnis, Alin Dobra and Sanjay Ranka. Analyzing the techniques that improve fault tolerance of aggregation trees in sensor networks. Journal of Parallel and Distributed Computing, Vol. 69 (12), pp. 950-960, 2009.
- Laukik Chitnis, Alin Dobra, and Sanjay Ranka, Fault tolerant aggregation in heterogeneous sensor networks. Journal of Parallel and Distributed Computing. Vol 69(2): 210-219 (2009)
- Laukik Chitnis, Alin Dobra and Sanjay Ranka. Aggregation Methods for Large Scale Sensor Networks. ACM Transactions on Sensor Networks, Volume 4 , Issue 2 (March 2008).
- Kannan Rajah, Sanjay Ranka, Ye Xia, Advance Reservation and Scheduling for Bulk Transfers in Research Networks, IEEE Transactions on Parallel and Distributed Systems, December 2008
- Kannan Rajah, Sanjay Ranka and Ye Xia, Scheduling Bulk File Transfer with Start and End Times, Scheduling Bulk File Transfers with Start and End Times, Computer Networks, Vol 52(5), April 2008, pp. 1105-1122.
- Eun-Sung Jung, Yan Li, Sanjay Ranka and Sartaj Sahni, An Evaluation of In-Advance Bandwidth Scheduling Algorithms for Connection-Oriented Networks, Proceedings of International Symposium on Parallel Architectures, Algorithms, and Networks, 2008, pp. 62-67.
- Eun-Sung Jung, Yan Li, Sanjay Ranka and Sartaj Sahni, Performance Evaluation of Routing and Wavelength Assignment Algorithms For Optical Networks, IEEE Symposium on Computers and Communications, 2008.
- Kannan Rajah, Sanjay Ranka and Ye Xia, Scheduling Bulk File Transfer with Start and End Times, Proceedings of NCA 2007, pp. 295-298.
- Zhe Wang, Sanjay Ranka and Ye Xia, Slotted Wavelength Scheduling for Bulk Transfers in Research Networks, 2009 International Conference on Parallel Processing to appear
- Sartaj Sahni, Nageshwara Rao, Sanjay Ranka, Yan Li, Eun-Sung Jung, Nara Kamath, Bandwidth Scheduling and Path Computation Algorithms for Connection-Oriented Networks, Proceedings of ICN 2007 (best paper award).