Annual Review of Scalable Computing

The following sections are included:

• Preface

• Contents

Recently, the demand in tool support (performance analyzers, debuggers etc.) for efficient Java programming considerably increases. A universal, open interface between tools and a monitoring system, On-line Monitoring Interface Specification (OMIS), and the OMIS compliant monitoring system (OCM) enable to specify such a Java oriented monitoring infrastructure which allows for an extensible range of functionality intended for supporting various kinds of tools. The paper presents an approach to building a monitoring system which underlies this infrastructure, i.e. the issues of architecture, request processing, event handling, and processing RMI calls.

The following sections are included:

• Motivation and Overview

• P2P Systems as Components for Scalable Grids

• A Scalable Data Grid with P2P Components

• Availability Component
  • Basic Equations and Model Parameters
  • Analytical Model
  • Applying the Model

• Synchronization Component

• Self-Management Components

• Putting it All Together

• Related Work

• Conclusion

• Acknowledgements

• Bibliography

In this paper we present a study on the subject of the Immediate Dependency Relation (IDR), and we show how by extending the IDR relation, one can ensure a global causal delivery in group communication, including in the overlapping group case. The main objective of this paper is to show that the use of the Immediate Dependency Relation (IDR) obliterates the notion that causality is expensive to set up in distributed systems. The IDR minimizes considerably the amount of control information sent per message to ensure causal ordering, without imposing restrictions in interaction (e.g. network topology, redifussion servers, executions models etc.). In order to demonstrate the feasibility of its implementation, we present an optimal broadcast causal protocol inspired by the IDR relation. We show the optimality of the protocol in terms of the amount of control information timestamped per message.

The following sections are included:

• Introduction

• Preliminaries
  • A Computing Model with Optical Buses
  • Scalability of Parallel Computations

• Limitations of Static Networks

• Basic Matrix Operations

• Matrix Multiplication
  • Bilinear Algorithms
  • Implementation Details
  • High Scalability

• Matrix Chain Product and Matrix Powers
  • Matrix Chain Product
  • Matrix Powers

• Other Matrix Computations
  • Inversion of Lower and Upper Triangular Matrices
  • Characteristic Polynomials, Determinants, and Ranks
  • Inversion of Arbitrary Matrices
  • Linear Systems of Equations and Krylov Matrix
  • LU- and QR-Factorizations

• Summary

• Acknowledgments

• Bibliography