Barry J. Brachman

Fragmentation in store-and-forward message transfer

The advantages of store-and-forward (S/F) message transfer at the application layer and problems with existing systems are discussed. Current systems suffer from the limitation that if an intermediate message transfer agent (MTA), the entity responsible for storing and forwarding messages, cannot store a message in its entirety due to insufficient buffer space, then the transfer fails. A mechanism is introduced, called message fragmentation, to transfer messages too large to be completely stored on an intermediate MTA automatically and efficiently. In addition to letting a large message pass through the network, fragmentation improves performance in an S/F message-based system. Schemes developed for flow control, deadlock prevention, and buffer management in packet-switched networks are described, and their suitability for use in the message-handling environment is discussed. Several novel schemes for buffer management in the message-handling environment are presented. The EAN X.400-based S/F message-handling system developed at the University of British Columbia is briefly cited as an example.<<ETX>>

Flow and congestion control in the message handling environment

A framework is presented for the design of application level store-and-forward message transfer systems. Flow control and congestion control issues in the message handling environment are discussed. A new model for message transport, the message stream, is proposed as a basis of the flow and congestion control functionality. Important design aspects, including message fragmentation and management of transit and recipient buffers, are addressed.<<ETX>>

A hierarchical solution for application level store-and-forward deadlock prevention

Application level message transfer systems, primarily used to transport electronic mail, have several unique characteristics that make using existing methods of dealing with deadlock impractical. The special requirements for the efficient transport of application level messages are discussed and a description of a hierarchical scheme that meets these requirements is given. For networks consisting of interconnected clusters of computers, the structured buffer pool technique can be combined with other deadlock-free message transport schemes to form a network-wide, deadlock-free message transport facility. A properly chosen hierarchical system has the advantages of the individual schemes while lessening the impact of their disadvantages. The hierarchical scheme is also applicable to packet-switched and datagram-based networks.

A simulation study of application level message transfer using message streams

Abstract Application level message transfer systems, used primarily to convey electronic mail, have several unique characteristics that have important performance-related consequences. A new approach to resource management in the message handling environment is outlined that addresses some of the weakness of current message transfer systems. Two important elements of this approach are the message stream and recipient buffer space management. A discrete-event simulation study was undertaken to evaluate the proposed mechanisms in comparison to a model representing current systems. Network throughput statistics were gathered for four simulated network topologies. End-to-end delay statistics and the influence of network load were studied in detail for one of them. The study indicates that the message stream approach to message transfer offers significantly higher throughput and has fairness advantages over current systems.

Resource management in application level message transfer systems

This thesis is concerned with the design of the resource management components of application level, store-and-forward message transfer systems. Although these systems have for some time been used to transport electronic mail, there has been little investigation into designs that emphasize performance and correctness aspects. Current message transfer systems are loosely structured in that there is little, if any, end-to-end resource management. The thesis beings by characterizing the message handling environment and comparing the message transfer approach to that of connection-based file transfer. Current message transfer systems have a fundamental limitation in that the largest message that can be transferred is determined by the amount of storage available at any of the intermediate hosts along the messages route. Major components of a message transfer system and design alternatives are discussed. Existing schemes that deal with solutions designed for lower networking levels are reviewed and shown to be inadequate in addressing the problems in the message handling environment. A framework for designing message transfer systems is presented. Systems adhering to the design methodology address performance issues in a structured way. Two new techniques are central to this framework: message fragmentation and the message stream. Message fragmentation is introduced as a means of delivering arbitrary size messages. The message stream abstraction is the basis of flow control and congestion control. A hierarchical technique for deadlock prevention in the message handling environment is introduced. In this method, the structured buffer pool approach is used as a top level and is integrated with a second method at the bottom level to produce a practical, deadlock-free message transfer system. Methods for providing transit buffer management and recipient buffer space allocation are discussed. A simulation study of some of the performance aspects of message streams and recipient buffer space allocation is presented.