Time, clocks, and the ordering of events in a distributed system
Communications of the ACM
A Customizable Simulator for Workstation Networks
IPPS '97 Proceedings of the 11th International Symposium on Parallel Processing
DieCast: testing distributed systems with an accurate scale model
NSDI'08 Proceedings of the 5th USENIX Symposium on Networked Systems Design and Implementation
Interface connecting the INET simulation framework with the real world
Proceedings of the 1st international conference on Simulation tools and techniques for communications, networks and systems & workshops
Integration of SCTP in the OMNeT++ simulation environment
Proceedings of the 1st international conference on Simulation tools and techniques for communications, networks and systems & workshops
Efficient and Scalable Network Emulation Using Adaptive Virtual Time
ICCCN '09 Proceedings of the 2009 Proceedings of 18th International Conference on Computer Communications and Networks
MPI-NeTSim: A Network Simulation Module for MPI
ICPADS '09 Proceedings of the 2009 15th International Conference on Parallel and Distributed Systems
Hi-index | 0.00 |
The coupling of real implementations with simulations helps the developer to better understand the behavior of the real system by shifting to the simulation the parts of the overall system that he cannot influence. This was done for MPI programs in MPI-NeTSim. where the network and transport protocol are simulated in OMNeT++ but the original MPI code executes unchanged. We previously developed a static time factoring algorithm to accurately handle the discrepancies between the wall clock time of the application and the virtual simulation time. However, the original algorithm proved to be too slow for simulations with a higher number of hosts and a greater amount of data. In this paper, we introduce a faster adaptive algorithm that no longer uses time factors nor requires a full system-wide synchronization but still meets the conditions of the ordering of events in a distributed system. The new algorithm scales to significantly more hosts and data.