We present a novel approach to building and deploying network protocols. The approach is based on mobile code, demand loading, and caching techniques. The architecture of our system allows new protocols to be dynamically deployed at both routers and end systems, without the need forcoordination and without unwanted interaction between co-existing protocols. In this paper, we describe our architecture and its realization in a prototype implementation. To demonstrate how to exploit our architecture, we present two simple protocols that operate within our prototype to introduce multicast and mobility services into a network that initially lacks them. 1

. Current paradigms for interprocess communication are not sufficient to describe the exchange of information at an adequate level of abstraction. They are either too low-level, or their implementations cannot meet performance requirements. As an alternative, we propose distributed shared objects as a unifying concept. These objects offer user-defined operations on shared state, but allow for efficient implementations through replication and distribution of state. In contrast to other object-based models, these implementation aspects are completely hidden from applications. 1 Introduction Communication can be viewed at different levels of abstraction. At a high level, it appears as an exchange of information between processes. These processes are either contained in a single parallel or distributed application, or may otherwise belong to different applications that need to communicate. At a low level, communication appears as the mere transfer of bits from one address space to another...

This paper describes the implementation and use of a prototype multicast late-binding RPC (LbRPC) system. The system provides a mechanism for exporting arbitrary code and data across a wide area internetwork to multiple hosts for evaluation. This mechanism provides a means to ensure that the total execution time for the application is as small as possible without requiring continual monitoring and estimation of the state of the hosts and the intervening network. The impact of sending the same code and data to a group of remote hosts is minimized by using multicast Internet protocol (IP) communications for the outward leg of the transaction. This prototype implementation demonstrates the feasibility of using multicasting for process placement and some performance measurements are included to show how it stands up to usage on the Internet. The paper concludes with a discussion of outstanding issues raised by this research

Introduction Distributed computing is inherently coupled to networking and telecommunications. As computing systems grow and the need for computing cycles increases to solve ever more complex problems the scaling advantage of distributed computing environments becomes more pronounced. Unfortunately as the diameter of wide-area networks become larger, latency becomes a formidable hurdle since communications channels are limited by the speed of light. This is because the completion time of a task is the time to complete some work and the amount of communication that must take place. The communication term is a function of both the number of times communication must occur, f(# messages), and the amount of data that must be exchanged, g(message). Latency has an effect on both f and g but bandwidth has a direct impact on g. If the amount of data to exchange is large, the

) Frederick Knabe November 2, 1994 Abstract New classes of distributed applications are appearing that require the ability to transmit executable code. To support these applications, we have implemented a function transmission facility for Facile, a distributed extension of Standard ML. Satisfying the needs of efficiency, heterogeneity, and distributed type safety required some modifications to the language. In this paper we report on the design choices we made and draw some conclusions about the high-level language features necessary to support function transmission. Word count: 4991 1 Introduction A common characteristic of distributed applications is that they exchange data over a network. Until recently, that data was generally limited to primitive types such as integers and strings along with various structured types. Now, though, a new class of distributed applications is beginning to appear that also relies on the ability to transmit code. Interest has been growing in appl...

Abstract – Mobile wireless clients have highly variable network connectivity and available bandwidth. There are times when they may be completely disconnected from the larger Internet. This dynamism of connectivity poses unique constraints for the problem of information access in general, and Web access in particular. These and other factors (such as loaded servers and congested networks) contribute to unpredictably high response times in content retrieval. We examine the problem of improving the utility of information access applications for these imperfectly connected devices. In particular, we are concerned with three related problems: 1. Providing information to the user on the estimated response time to retrieve content, the freshness of cached content, and the status on the strength of connection to the

It is widely believed that there is a lot of spare capacity -- CPU, memory and disk -- within networks of high-performance workstations. There have been a large number of surveys of lans, clusters and portions of the Internet to measure either utilisation or spare capacity. They have used traditional unicast polling or beaconing methods to measure a set of variables which charactise some aspect of machine utilisation. We have developed a tool which enables observation of the behaviour of networks of workstations. We use IP multicast and the experimental mbone to propagate the performance information to those who want to receive it. Multicast gives the tool a scalable architecture which does not consume a great deal of network or CPU resources. We describe the design and implementation of these distributed load monitoring and analysis tools which are based on the notion of local analysis of globally available data. This work is supported by JISC/NTI 229, "Distributed SuperComputing: H...

Abstract – Mobile wireless clients have highly variable network connectivity and available bandwidth. There are times when they may be completely disconnected from the larger Internet. This dynamism of connectivity poses unique constraints for the problem of information access in general, and Web access in particular. These and other factors (such as loaded servers and congested networks) contribute to unpredictably high response times in content retrieval. We examine the problem of improving the utility of information access applications for these imperfectly connected devices. In particular, we are concerned with three related problems: 1. Providing information to the user on the estimated response time to retrieve content, the freshness of cached content, and the status on the strength of connection to the