Sub Ramakrishnan

A close look at task assignment in distributed systems

C. C. Shen and W. H. Tsai (IEEE Trans. Comput., vol.C-34, no.3, p.197-203 1985) proposed a graph matching algorithm for solving the static task assignment problem. It combines two important ideas: (1) graph homomorphism and (2) application of the A* algorithm. Task-dependent information is used as a heuristic to reduce the search effort in finding an optimal path to the goal node. An examination is made of Shen and Tsais strategy and their complexity measure. The authors propose some simple alternatives to their algorithm that are effective in reducing the number of nodes generated (and expanded) without sacrificing the optimality criteria.<<ETX>>

A genetic approach to color image compression

Advances in compression technique will have to keep pace with the exponential increase in the need for storage and transport of bulky multimedia images. This paper concerns the use of vector quantization for processing color images. We propose a new, genetic algorithm based, scheme for code book design in the quantization process, and compare its performance with the widely used LindeBuzo-Gray algorithm [8]. We also introduce a new genetic operator known as synchronization that works especially well in this problem domain. The quality of the resulting code book is used as a performance criteria for LBG and for the proposed approach. Our observations reveal that the proposed scheme is better than the LBG algorithm by a factor of between 5% to 25%. The performance gain is especially significant for large code books.

Classroom projects on database connectivity and the Web

Most undergraduate curriculums in CS include at least a one semester study of data base management systems. In this paper, we take a closer look at a DBMS course we have offered recently and elaborate on a world wide web based database project in this course; it integrates a number of ideas not only from within DBMS but across other CS areas. We discuss the objectives underlying the design of the project and the ways in which these projects provide students with a feel for the organization and use of contemporary commercial database applications.

Models for performance evaluation of a local area network with a file server

Abstract We consider a system of single user workstations connected to a file server over a local network. Users arrive at random intervals and occupy a workstation if one is available otherwise leave the system. Upon arrival, the user issues a request for file transfer from the server to the workstation local memory. Subsequent requests for additional transfers are issued in a random manner as needed. We consider two priority schemes for scheduling the two types of requests at the file server. Assuming that all relevant intervals are exponentially distributed, we propose two queueing models for the two scheduling schemes and develop computational procedures to analyze the performance of the system. Illustrative examples are presented to study the system behavior under various load conditions and to compare the performance of the two scheduling disciplines.

DBMS course: web based database administration tool and class projects

In this paper, we discuss a software tool we have developed for use in undergraduate DBMS courses, that provides: (i) a web-enabled database set up and administration facility for faculty use, and (ii) web-based database projects that may be assigned to students.We discuss the motivation for our work and the objectives underlying the design of the tool. The significance of our work is two-fold. First, it should help the instructor in the set up, management, and monitoring of student database accounts. Second, it provides the students with a feel for the organization and use of contemporary web-enabled database applications.

A heuristic cost estimation method for optimizing assignment of tasks to processors

This paper concerns optimal static assignment of a group of communicating tasks among a set of processors. The assignment is based on Shen and Tsai’s [I] criteria for optimizing the completion time (or delay) of the task group. The delay criteria enhances parallelism by spreading the computation or load across multiple processors. In this paper we discuss the application of a heuristic cost estimation method that results in a speed up in solving the delay optimization problem. We use an example to demonstrate our approach and make observations based on running a large number of data sets using the proposed scheme.

Java based e-commerce middleware

Organizations that sell products over the Internet usually have a product database located at or near the web hosting server. Web clients use SSL to connect to the web server and order products, which are then entered into the product database. Since web hosting services may be outsourced, the order information needs to be transmitted securely to the corporate database of the enterprise that sells these products. This paper concerns the development of a secure middleware application in Java that connects the web hosting database with the corporate backend database [9]. We discuss the design details of the middleware and implementation issues of some of the software pieces of this puzzle. 1 Background

Extension of Stone's task assignment model for distributed computer systems

The authors consider optimal static assignment of a group of communicating tasks among a set of processors. The assignment is based on the throughput metric optimality criteria by H.S. Stone (1977) which is the total cost for execution and interprocess communication. Stones model is generalized with the addition of an important factor called processor connectivity. The scheme is necessary in situations where the processors need not exhibit full connectivity. It is shown that a graph-theoretic approach similar to that proposed by C.C. Shen and W.H. Tsai (1985) for a delay metric can be used to solve the ensuing problem and it is found empirically that the procedure is computationally more efficient for the throughput metric.<<ETX>>

An integrated optimal task assignment policy

Given a set of tasks the task assignment problem ia to determine the processor from a processorset at which a task can reside for its lifetime. Each task has a known execution cost on each of the processors, and may cmnrnunicatc with zero or more other tasks in the task set. ‘lko ccmrmmicating tasks incur an intmpmceas communicationoverheadwhen they are assigned to two different processors. We propose an algorithm for task assignment problem which finds optimal solutions. The algorithm has exponential complexity; however, it performs quite well in practice in comparison to previous methods, and can be applied in a practical environment. Our assignment is baaed on Stone’s [13] optimti~ criteria which is the total cost for execution and interpcess communication.

A protocol for program migration

The authors address the problem of remote computing in a disturbed environment and describe the features of a simple hybrid protocol for program migration. Simulations indicate that the number of overhead messages can be reduced considerably. The algorithm uses a hybrid scheme (a combination of sender-initiated and receiver-initiated). The first approach is used when the system is heavily loaded, and the second when the system is lightly loaded. The system toggles between these two schemes on a dynamic basis. The algorithm exhibits fairness and collapse of one node does not lead to total system collapse.<<ETX>>