[GRID COMPUTING]

Topics of interest:

Design of application specific Cluster/Grid computing systems/algorithms; Algorithms for handling Large-Scale Data on Grids; Resource Allocation; Load Balancing; Pricing on Grid Systems; Media-Grid (Interactive Digital Media Support)

Scroll down for the details of funded projects.

Something here on what I am doing!

This is a very vast domain and literally no boundaries to confine! I am a researcher with a background in Parallel & Distributed Computing, specifically, in the area of Scheduling. Grid Computing seems to offer considerable challenge in handling Scheduling Problems. This mainly due to:

· Heterogeneity  of the infrastructure – No defined topology, a crude mix of several types of processing components with different capabilities;

· Availability – Take an instance CONDOR like systems wherein nodes enter and leave arbitrarily thus making the problems all the more complex as far as scheduleability decisions are to be made!

· Resource Management – With the above two characteristics, managing resources (CPUs, memory availability at nodes, link availabilities for establishing the desired connections, brokering, negotiations kind of dealings, resource discovery, etc) becomes extremely complex.

My interests, as pointed in this area, addresses and encompasses the above three issues, in a variety of forms.  For example, load balancing problem for Grid infrastructures strictly cannot make use of conventional load balancing algorithms designed for distributed systems. However, certain principles can be borrowed for an easy start in Grid Computing domain! As Grid’s horizon cannot be restricted, traversing across such a monster network is a communication costly task! Especially, the delays can be fluctuating across nodes and availability of paths across a domain is questionable!

To Students and beginners: There are number of theoretical challenges awaiting in this domain! For those who aspire to work in Optimization area, this will be your home-ground! Tools such as Linear Algebra, Calculus, Real Analysis fundamentals, Probability theory are essential to make yourself comfortable in this domain!

I work in addressing above three issues and these are fundamental problems akin to Grid. If those issues are not addressed carefully, use of Grid infrastructure becomes meaningless and defeats the entire purpose of migrating to a Grid. These issues become very sensitive, when handling large-scale computational loads on Grids!

NOTE - For those who just wish to tap enormous power available on Grids for their computational purposes [bioinformatics, astronomical studies, physics, complex mathematical computations, etc], all that you need to know is on how to submit your jobs to a “Grid-enabled node”! It is highly transparent in that sense!

============================================

ABOUT MY FUNDED PROJECT:

(Current External Funding) [Grid Computing] Received a competitive external academic research grant from A*STAR SERC (through National Grid Office, Singapore) in July 2005 for ~USD 182,000.00(~S$300K). (Title: Design of Resource Management and large-Scale Data Processing Strategies for Grid Computing Environments).

START DATE - 20/09/2005; END DATE – 19/09/2007

I am the Principal Investigator of this project looking at issues related to my above description of the three characteristics of Grid Computing domain. Other co-PIs are investigating on communication and pricing aspects.

BRIEF DESRIPTION OF THE PROJECT (but certainly sufficient to understand what is going on!)

The objectives and directions of the project are multi-fold. The project primarily aims to design efficient resource aware strategies that optimize the performance of a grid computing environment (GCE) under several influencing factors. The project addresses several critical issues that include, design of resource aware grid job dispatcher for efficient scheduling, QoS issues on underlying communication networks, resource management strategies, large scale load distribution algorithms, etc. In this proposal we take a radically different approach in devising efficient techniques that optimize several performance metrics such as, job throughputs, quality of service parameters, resource utilization, response time, queue completion time at the scheduler and grid nodes, respectively. The algorithms to be designed are suitable and  easily adaptable to a wide range of application domains such as, data processing in high-energy particle physics experiments, modeling epidemic spread and control, etc, to quote a few. Below, we list the objectives, which subsume deliverables of this project. Detailed list of objectives can be found in Case for Support attached.

Objectives and Deliverables:

· To design resource aware scheduling and resource sharing schemes for a GCE;

· To develop load balancing algorithms across privileged grid-enabled sites;

· To develop and deploy a joint scheduling strategy handling multiple loads from multiple grid sites;

· To develop efficient mechanisms to dynamically discover resources, provision bandwidth-guaranteed tunnels, schedule sub-wavelength and wavelength-circuits

· To develop and simulate algorithms to ensure QoS requirements of services and applications are satisfied at both application and network levels;

· To design and deploy a resource management and scheduling information and control architecture in an actual grid computing architecture, e.g. Globus, Gridbus;

· To evaluate the effectiveness of the strategies by rigorous simulation studies using real-life parameters and modeling for high energy particle physics experiments;

· To make recommendations on the design, usefulness, and applicability of the results from this proposal.

Achieving the above set of objectives and deliverables will lead to a significant improvement in the efficiency of resource allocation (thus, enabling a larger number of users and applications to be served), as well as the achievable performance of compute-and-data (communication and storage) applications on the grid.

============================================

Published or Accepted Works in Grid Domain: [We respect the copyright issues and hence, we are not releasing the soft-copies of the papers. I may be contacted for off-prints! As some of these works are now in-press, I can send a Technical Report of these works. Please do not forget to cite if you are using these works in any form! Thanks.]

Books/Chapters:

· [Invited Chapter] Bharadwaj Veeravalli "Designing High-Performance Concurrent Strategies for Biological Sequence Alignment Problems on Networked Computing Platforms", In an edited Book by El-Ghazali Talbi and A. Zomaya on Grids for Bioinformatics and Computational Biology, Wiley & Sons, USA, 2007.

· [Book Chapter] Ruchir S., Bharadwaj V., and, Manoj M., "Estimation Based Load Balancing Algorithm for Data-Intensive Heterogeneous Grid Environments", To appear as a Chapter in Lecture Notes in Computer Science (LNCS), Springer-Verlag, 2006.

Journals:

· Benjamin Khoo Boon T, Bharadwaj Veeravalli, Terence Hung, Simon See, "A Multi-Dimensional Scheduling Scheme in a Grid Computing Environment", To appear in Journal of Parallel and Distributed Computing (JPDC), 2007.

· Ruchir S, Bharadwaj, V, and Manoj, M, "On the Design of Adaptive and De-centralized Load Balancing Algorithms with Load Estimation for Computational Grid Environments", To appear in IEEE Transactions on Parallel and Distributed Systems, 2007.

· Sivakumar Viswanathan, Bharadwaj Veeravalli, and Thomas. G. Robertazi, "Resource Aware Distributed Scheduling Strategies for Large-Scale Computational Cluster/Grid Systems", To appear in IEEE Transactions on Parallel and Distributed Systems, 2007.

· Darin England, Bharadwaj Veeravalli, and Jon Weissman, "A Robust Spanning Tree Topology for Data Collection and Dissemination in Distributed Environments", To appear in IEEE Transactions on Parallel and Distributed Systems, 2007.

· Benjamin Khoo and Bharadwaj Veeravalli, "Cluster Computing and Grid 2005 Works in Progress: A Dynamic Estimation Scheme for Fault-Free Scheduling in Grid Systems ", IEEE Distributed Systems Online, vol. 6, no. 9, 2005.

Conferences:

· Ruchir S., Bharadwaj V., and, Manoj M., "Estimation Based Load Balancing Algorithm for Data-Intensive Heterogeneous Grid Environments", In the Proceedings of the 13th IEEE International Conference on High Performance Computing (HiPC 2006), Bangalore, India, pp. 72-83, 2006.  

· Qin Z, Bharadwaj V, and Tham Ck, "Fault-tolerant Scheduling for Differentiated Classes of Tasks with Low Replication Cost in Computational Grids", To appear in  the proceedings of HPDC 2007, Monterey Bay California, USA.

· Ruchir S., Bharadwaj V., and, Manoj M., "A Receiver-initiated Load Balancing Algorithm with Parameter Estimation for Heterogeneous Grid Environments", To appear in the Proceedings of the 2nd International Conference on Semantics, Knowledge and Grid (SKG-2006), IEEE CS Press, 2006.

· Benjamin Khoo Boon Tat, Bharadwaj Veeravalli, Terence Hung, and Simon See Chong Wee, "A Co-ordinate Based Resource Allocation Strategy for Grid Environments", In the proceedings of  6th IEEE International Symposium on Cluster Computing and Grid (CCGRID) 2006, Singapore, 16-19 May, pp561-567, 2006. [26% acceptance rate]

· S. Viswanathan, B. Veeravalli, D. Yu, T.G. Robertazzi, "Design and Analysis of a Dynamic Scheduling Strategy with Resource Estimation for Large-Scale Grid Systems", In the Proceedings of the 5th IEEE/ACM International Workshop on Grid Computing (held in Conjunction with SuperComputing 2004), Pittsburgh, Pennsylvania, USA, pp. 163-171, November 2004.

· Wong Han Min, Bharadwaj Veeravalli, Danyong Yu, and T. G. Robertazzi, "Data Intensive Grid Scheduling: Multiple Sources with Capacity Constraints", In the Proceedings of the International Conference on Parallel and Distributed Computing Systems (PDCS), USA, 2003.