Benchmarking of hard real-time distributed systems with Ada 95
IRTAW '97 Proceedings of the eighth international workshop on Real-Time Ada
IPDPS '05 Proceedings of the 19th IEEE International Parallel and Distributed Processing Symposium (IPDPS'05) - Workshop 15 - Volume 16
A Real-Time Performance Evaluation Model for Distributed Software with Reliability Constrains
The Journal of Supercomputing
CIS'04 Proceedings of the First international conference on Computational and Information Science
Hi-index | 0.00 |
Practically, there are endless variations of the implementation of real-time distributed systems (RTDSs). These variations are the consequences of the different hardware, system software architectures and application requirements. These variations make the testing of RTDSs more implementation dependent. To the contrary benchmarking can be considered to be domain specific. In general any RTDS can be considered to have three scheduling domains: the node's processor scheduling domain, the node's communication scheduling domain, and the nodes' priority domain. The only way a benchmark would be able to integrate these three scheduling domains and also be hardware and system software architecture independent, is to present itself as an application to the RTDS. From the application point of view, the RTDS can be seen as set of tasks (processes) located on different nodes, which communicate with each other by exchanging messages. The tasks as well the messages have timing constraints. The Hartstone Distributed Benchmark (HDB) can be viewed as an application-oriented benchmark applicable to any type of RTDS. The HDB provides figures of merit for the performance evaluation of the end-to-end scheduling of messages.