Hybrid Packet/Fluid Flow Network Simulation
Proceedings of the seventeenth workshop on Parallel and distributed simulation
Performance Benchmark of a Parallel and Distributed Network Simulator
Proceedings of the 19th Workshop on Principles of Advanced and Distributed Simulation
Packet-level integration of fluid TCP models in real-time network simulation
Proceedings of the 38th conference on Winter simulation
Hybrid simulation of a FIFO queuing system with trace-driven background traffic
Proceedings of the 2nd international conference on Performance evaluation methodologies and tools
An agent-supported simulation framework for metric-aware dynamic fidelity modeling
SpringSim '07 Proceedings of the 2007 spring simulation multiconference - Volume 2
A fluid background traffic model
ICC'09 Proceedings of the 2009 IEEE international conference on Communications
Energy efficiency for large-scale MapReduce workloads with significant interactive analysis
Proceedings of the 7th ACM european conference on Computer Systems
| Hi-index | 0.00 |
A number of methods exist that can be used to create simulationmodels for measuring the performance of computernetworks. The most commonly used method ispacket level simulation, which models the detailed behaviorof every packet in the network, and results in ahighly accurate picture of overall network behavior. Aless frequently used, but sometimes more computationallyefficient, method is the fluid model approach. In thismethod, aggregations of flows are modeled as fluid flowingthrough pipes, and queues are modeled as fixed capacitybuckets. The buckets are connected via pipes, wherethe maximum allowable flow rate of fluid in the pipes representsthe bandwidth of the communication links beingmodeled. Fluid models generally result in a less accuratepicture of the network's behavior since they rely on aggregationof flows and ignore actions specific to individualflows.We introduce a new hybrid simulation environment thatleverages the strong points of each of these two modelingmethods. Our hybrid method uses fluid models to representaggregations of flows for which less detail is required,and packet models to represent individual flows for whichmore detail is needed. The result is a computationally efficientsimulation model that results in a high level of accuracyand detail in some of the flows, while abstractingaway details of other flows. We show a computationalspeedup of more than twenty in some cases, with littlereduction in accuracy of the simulation results.