Herbert Eberle

Generic Support for Distributed Processing in Heterogeneous Networks

The apparent complexity of distributed application development, especially in a heterogeneous environment, is the prime motivation for the network operating system kernel described in this paper. The kernel reduces this complexity by separating the distribution related issues from the application related ones. It provides an interface of generic objects and operations, which are able to take away from the application programmer most of the problems of distribution, access protection, resource management, and data representation. This paper develops the basic concepts of the kernel from rather general design objectives and illustrates its use and major properties. A prototype implementation, which is running on three different architectures, demonstrates the feasibility of adding these facilities to a given operating system without affecting existing interfaces or applications. The paper reports about early experience with the implementation and performance of the prototype.

An architecture for the cooperation of heterogeneous operating systems

The DAC Network Operating System (DACNOS) was designed to support resource sharing in a world of interconnected heterogeneous computing systems. The prototype implementation demonstrates that it is feasible to add powerful and flexible means for distributed cooperation to an operating system without affecting its existing individual interfaces and applications. The authors describe the important design issues of DACNOS and their experiences with the implementation and performance of a prototype. Particular emphasis is put on the portability of the NOS software and on the design of the interface to the NOS kernel that provides the facilities for distributed cooperation. It is shown by an example how this set of facilities eases the implementation of distributed applications by taking most of the burden of distribution, access protection, resource management and data representation away from the programmer.<<ETX>>

Operating System Support for Heterogeneous Distributed Systems

Our view of a distributed system is a set of cooperating autonomous nodes interconnected by a fast communication network. Each node has some processing and storage capabilities. There is no physically shared memory between nodes. All communication is accomplished via some network transport service.

Network Operating System Kernels for Heterogenous Environments

A major function of a Network Operating System Kernel is to free application program developers from the need for designing complex protocols. The Operating System analogy for networks offers a paradigm for inter-program cooperation, as if this cooperation were local. Instead of forcing programs to send and interpret messages explicitly, the Network Operating System provides services for remote operations on accustomed data structures, whereby the message exchange as such remains invisible to the programmer.

Extending DCE RPC by Dynamic Objects and Dynamic Typing

Current DCE RPC offers an object model which is static in that objects are assumed permanent and published within a directory. Moreover, all parameters of an objects interface are typed statically. We argue that this object model should be enhanced by dynamic objects which are created as the result of a client/server interaction with parameters whose types are conveyed at creation- or call-time. Dynamic objects are a common model for context-handles and callback, as well as delegation scenarios. Dynamic typing facilitates access to generic servers through RPC. In this paper our emphasis is on motivating the need for the proposed extensions, and on showing that they can be integrated into current DCE RPC in an upward compatible manner.

Virtual machine dispatching under fairness constraints

The paper analyzes the problems of low level dispatching under predictability constraints, with good responsiveness and low overhead. It is shown that nonpreemptive and/or round-robin disciplines fail for the general case. A novel algorithm is proposed based on sorting at time slice intervals. Simulation results are used to discuss the properties of the new dispatching algorithm.

Distributed academic computing

The main objective of the DAC project is to contribute to the development of a systematic approach to resource sharing in a heterogeneous network of autonomous computers. Ideally, users should have access to remote resources from their own environment independent of the location of the resource. Owners of resources should maintain full control over the use of their resources and resource allocation should be accomplished according to their objectives. Users should be protected from misbehavior or malice of other users. Important additional design objectives are compatibility and portability: the native user interfaces of local operating systems should remain unchanged, and access to remote resources, wherever possible, should be provided through the native user interfaces; further, it should be easy to port the solution to different types of operating systems, network architectures and computer architectures.