System model and taxonomy for host-network interfaces
Date of Award
Doctor of Philosophy (PhD)
Electrical Engineering and Computer Science
Geoffrey C. Fox
ATM, Networks, Taxonomy, Host-network interfaces
Computer and Systems Architecture | Computer Sciences | OS and Networks
The performance of parallel and distributed applications running on network of workstation resources suffers from high latency and low throughput even when the network speed is in the Giga bit/second range. The main reasons for the low performance are due to the design of the host-network interface and the software mechanisms that are used to transmit data from the host to the network, and vice versa.
In this dissertation, we have reviewed host-network interface designs and propose a comprehensive taxonomy to characterize and describe the architectures of host network interface. In the architectural context, we have described the main components used to implement the host-network interface functions and how they interact with each other. In the protocol context, we have discussed how the software functions (data copying, packetization, error handling, flow control, routing) are mapped into the main host-network-interface components identified at the architecture level.
We have also provided several examples that show how our taxonomy can accurately describe the architectures of host-network interface and characterize the performance limitations of any host-network interface designs as well.
We support the design of host-network interface systems with the quantitative performance model, which is derived from the comprehensive taxonomy. Above all, our designs in host-network interfaces are motivated by a desire for increased performance, so it is instructive at this point to consider how the host-network interface system could achieve better performance.
Surface provides description only. Full text is available to ProQuest subscribers. Ask your Librarian for assistance.
Na, Bokyun, "System model and taxonomy for host-network interfaces" (2000). Electrical Engineering and Computer Science - Dissertations. 155.