A performance analysis model for distributed simulations

Distributed simulation (DS) aims at the acceleration of discrete event simulation experiments by decomposing the simulation model into spatially separated event sets, which are then executed in a distributed environment via synchronized logical processes (LPs). Synchronization protocols like Time Warp and Chandy/Misra/Bryant are employed to preserve global causality constraints and temporal consistency among events that occur locally within LPs. The performance of such LP based DS is influenced by a variety of factors such as the event structure underlying in the simulation model, the partitioning into submodels, the performance characteristics of the execution platform, the implementation of the simulation engine and optimizations related to the protocols. The mutual performance effects of parameters exhibit a prohibitively complex degree of interweaving, so that analytical performance models are of relative importance only. Nevertheless is a DS performance diagnosis of utmost practical interest when it comes to decide on the suitability of a certain DS protocol for a specific simulation model before substantial efforts are invested in developing sophisticated DS codes.

2005

In this work we illustrate the design and implementation guidelines of a recently developed middleware defined to support the parallel and distributed simulation of large scale, complex and dynamically interacting system models. The distributed simulation of complex system models, may suffer the communication and synchronization required to maintain the causality constraints between distributed model components. We designed and implemented the ARTÌS middleware as a new framework by incorporating a set of features that allow adaptive optimization by exploiting many complex and dynamic model and distributed simulation characteristics. As an example, a dynamic migration mechanism for the run-time adaptive allocation of model entities has been designed and exploited for dynamic load and communication balancing. Optimizations have been introduced to obtain the maximum advantage from heterogeneous and asymmetric communication systems, from shared memory to LAN and Internet communication. Other optimizations have been introduced by the exploitation of concurrent replications of parallel and distributed simulations, in order to increase the resources utilization and to maximize the speedup of simulation processes. Solutions have been designed, implemented and tuned to obtain a significant reduction in the communication and synchronization overheads between the physical execution units, and an increased model scalability and simulation speedup, even in worst-case modeling assumptions and simulation scenarios. Tools for Performance Evaluation of Complex Systems (FIRB-PERF'05) Recently implemented middleware solutions based on the High Level Architecture (HLA) [11, 20] have shown that the parallel and distributed simulation of massive and complex systems can result in relevant overheads. Overheads are due to the complex and full management of a wide set of run-time services and to the latency due to distributed communication bottlenecks. Specifically, most of the preliminary implementations of the interprocess communication services have been implemented in suboptimal way, without considering the heterogeneity of the simulation execution platforms [3, 13].

ArXiv, 2019

Simulation is a fundamental research tool in the computer architecture field. These kinds of tools enable the exploration and evaluation of architectural proposals capturing the most relevant aspects of the highly complex systems under study. Many state-of-the-art simulation tools focus on single-system scenarios, but the scalability required by trending applications has shifted towards distributed computing systems integrated via complex software stacks. Web services with client-server architectures or distributed storage and processing of scale-out data analytics (Big Data) are among the main exponents. The complete simulation of a distributed computer system is the appropriate methodology to conduct accurate evaluations. Unfortunately, this methodology could have a significant impact on the already large computational effort derived from detailed simulation. In this work, we conduct a set of experiments to evaluate this accuracy/cost tradeoff. We measure the error made if client-...

1998

Distributed simulation is generally not considered as an option when companies are deciding about their simulation methods. This might be due to the lack of distributed simulation software available or the complexity of implementing a distributed simulator. This paper presents aspects which should be considered when planning a distributed simulator and an experimental distributed simulator, Diworse, are presented. An example of distributed cellular system simulation illustrates that good speedups can be achieved when all aspects are properly considered.

Proceedings of the 19-th International Symposium on Computer and Information Sciences (ISCIS 2004), 2004

This paper illustrates the motivation, the preliminary design and implementation issues, of a new distributed simulation middleware named Advanced RTI System (ARTIS). The aim of the ARTIS middleware is to support parallel and distributed simulations of complex systems characterized by heterogeneous and distributed model components. The ARTIS design is oriented to support the model components heterogeneity, distribution and reuse, and to increase the simulation performances, scalability and speedup, in parallel and distributed simulation scenarios. Another design issue of the ARTIS framework is the dynamic adaptation of the interprocess communication layer to the heterogeneous communication support of different simulation scenarios. In this paper we illustrate the guidelines and architecture that we considered in the design and implementation of the ARTIS middleware, and we sketch some case studies that demonstrated the ARTIS utility and motivation, e.g., a distributed simulation of massively populated wireless ad hoc and sensor networks.

Proceedings of the 19th ACM/IEEE/SCS Workshop on Principles of Advanced and Distributed Simulation (PADS'05), 2005

Parallel and distributed simulations enable the analysis of complex systems by concurrently exploiting the aggregate computation power and memory of clusters of execution units. In this paper we investigate a new direction for increasing both the speedup of a simulation process and the utilization of computation and communication resources. Many simulation-based investigations require to collect independent observations for a correct and significant statistical analysis of results. The execution of many independent parallel or distributed simulation runs may suffer the speedup reduction due to rollbacks under the optimistic approach, and due to idle CPU times originated by synchronization and communication bottlenecks under the conservative approach. We present a parallel and distributed simulation framework supporting concurrent replication of parallel and distributed simulations (CR-PADS), as an alternative to the execution of a linear sequence of multiple parallel or distributed simulation runs. Results obtained from tests executed under variable scenarios show that speedup and resource utilization gains could be obtained by adopting the proposed replication approach in addition to the pure parallel and distributed simulation.

2011

Distributed computing has many opportunities for Modeling and Simulation (M&S). Grid computing approaches have been developed that can use multiple computers to reduce the processing time of an application. In terms of M&S this means simulations can be run very quickly by distributing individual runs over locally or remotely available computing resources. Distributed simulation techniques allow us to link together models over a network enabling the creation of large models and/or models that could not be developed due to data sharing or model reuse problems. Using real-world examples, this advanced tutorial discusses how both approaches can be used to benefit M&S researchers and practitioners alike.

IIE Transactions, 2001

With recent advances in parallel computation, distributed simulation has become a viable way of dealing with time-consuming simulations. For distributed simulations to run eciently, care must be taken in assigning the tasks (work) in the simulated system to the available physical processors in the computer system. An inecient assignment can result in excessive communication times between processors and unfavorable load conditions. This leads to long run times, possibly giving performance worse than that with a uniprocessor sequential event-list implementation. This paper establishes the feasibility, and in some cases the necessity, of using dynamic task allocation (rather than a-priori static allocation) in distributed simulation. A dynamic reallocation strategy is developed, and experiments on an iPSC/2 Hypercube indicate that signi®cant improvements in run time can be achieved at low cost.

IEEE Transactions on Instrumentation and Measurement, 2011

Communication in distributed simulations is highly relevant due to latencies cumulatively introduced in the execution time of such simulations; these latencies are caused by the transfer of data throughout networks. High-Level Architecture (HLA) is a well-known standard used to organize and keep the consistency of distributed virtual simulations. Nevertheless, although the HLA framework presents solutions to decrease communication overhead for simulations, it does not solve the communication issues that are pertinent to the network distances among federates and the Run-Time Infrastructure (RTI) in large-scale environments. Due to the relevance of load balancing for distributed simulations, many solutions have been proposed, but such approaches just consider communication load partially in their schemes or are limited. To minimize the communication overhead of HLA distributed simulations, a hierarchical dynamic balancing scheme composed of three phases was designed; however, even though the balancing scheme reacts properly to run-time load changes, its load analysis does detect all the imbalances after measuring simulations' communication load. Thus, an extension is proposed to improve the detection of communication imbalances, as well as the redistribution of federates, so the balancing scheme can better react to communication load measurements gathered from each federate. To observe the benefits of the proposed communication balancing scheme and its extension, extensive large-scale experiments were realized. The results show that the scheme reduces considerably the amount of communication overhead, and the extension enables the detection of communication imbalances by modifying the decision-making technique in the balancing system.

ACM Transactions on Modeling and Computer Simulation, 2013

Parallel and distributed simulations (PADS) realize the distributed execution of a simulation system over multiple physical resources. To realize the execution of PADS, different simulation infrastructures such as HLA, DIS and TENA have been defined. Recently, the Distributed Simulation Engineering and Execution Process (DSEEP) that supports the mapping of the simulations on the infrastructures has been defined. An important recommended task in DSEEP is the evaluation of the performance of the simulation systems at the design phase. In general, the performance of a simulation is largely influenced by the allocation of member applications to the resources. Usually, the deployment of the applications to the resources can be done in many different ways. DSEEP does not provide a concrete approach for evaluating the deployment alternatives. Moreover, current approaches that can be used for realizing various DSEEP activities do not yet provide adequate support for this purpose. We provide...