A Simulation Model for Large Scale Distributed Systems

Author(s):  
Ciprian Dobre

The use of discrete-event simulators in the design and development of Large Scale Distributed Systems (LSDSs) is appealing due to their efficiency and scalability. Their core abstractions of process and event map neatly to the components and interactions of modern-day distributed systems and allow designing realistic simulation scenarios. MONARC 2, a multithreaded, process oriented simulation framework designed for modeling LSDSs, allows the realistic simulation of a wide-range of distributed system technologies, with respect to their specific components and characteristics. This chapter presents the design characteristics of the simulation model proposed in MONARC 2. It starts by first analyzing existing work, outlining the key decision points taken in the design of the MONARC’s simulation model. The model includes the necessary components to describe various actual distributed system technologies and provides the mechanisms to describe concurrent network traffic, evaluate different strategies in data replication, and analyze job scheduling procedures.

2006 ◽  
Vol 14 (2) ◽  
pp. 151-170 ◽  
Author(s):  
Sharon Simmons ◽  
Dennis Edwards ◽  
Phil Kearns

Capturing and examining the causal and concurrent relationships of a distributed system is essential to a wide range of distributed systems applications. Many approaches to gathering this information rely on trace files of executions. The information obtained through tracing is limited to those executions observed. We present a methodology that analyzes the source code of the distributed system. Our analysis considers each process's source code and produces a single comprehensive graph of the system's possible behaviors. The graph, termed the partial order graph (POG), uniquely represents each possible partial order of the system. Causal and concurrent relationships can be extracted relative either to a particular partial order, which is synonymous to a single execution, or to a collection of partial orders. The graph provides a means of reasoning about the system in terms of relationships that will definitely occur, may possible occur, and will never occur. Distributed assert statements provide a means to monitor distributed system executions. By constructing thePOGprior to system execution, the causality information provided by thePOGenables run-time evaluation of the assert statement without relying on traces or addition messages.


2019 ◽  
Vol 37 (15_suppl) ◽  
pp. e18061-e18061
Author(s):  
Hui-Li Wong ◽  
Koen Degeling ◽  
Azim Jalali ◽  
Jeremy David Shapiro ◽  
Suzanne Kosmider ◽  
...  

e18061 Background: The wide range of possible combinations and sequences available for mCRC treatment presents a major challenge to clinicians, who need to determine the optimal approach for an individual patient or patient subset. In the absence of clinical trial evidence, real world data are an increasingly valuable resource that can be utilized not only to understand treatment patterns and outcomes in routine practice, but also to define an optimal treatment strategy for individual patients across multiple lines of therapy. Methods: Real world data from an Australian mCRC registry were used to develop an interactive data visualization tool that displays treatment variation, customizable to different levels of detail and specific patient subsets, based on patient and disease characteristics. Next, a discrete event simulation model was developed to predict progression-free (PFS) and overall survival (OS) for first line palliative treatment with doublet chemotherapy alone or with bevacizumab, based on data of 867 patients that were treated accordingly. Results: Of 2694 Australian patients enrolled, 2057 (76%) started 1st line treatment with chemotherapy and/or a biologic agent, 1087 (40%) and 428 (16%) received 2nd and 3rd line therapy, respectively. Combined, these 3 lines of treatment accounted for 733 unique sequences. After recoding treatment to the most intensive chemotherapy and the first exposed biologic, 472 unique sequences remained. In exploratory analyses, the simulation model estimated that median 1st line PFS (95% CI) of 219 (25%) patients could be improved from 175 (156, 199) to 269 days (247, 293) by targeting a different treatment. Conclusions: This was an initial exploration of the potential for data visualization and simulation modeling to inform understanding of practice in mCRC and to guide clinical decision making. Such tools allow clinicians and health system providers to define variation in practice patterns and to identify opportunities to improve care and outcomes. Ultimately, the aim is to improve the delivery of personalized cancer care, where other applications such as conditional survival and cost-effectiveness analyses may be useful.


2020 ◽  
pp. 204388692093702
Author(s):  
Damian Kedziora ◽  
Esko Penttinen

The teaching case addresses the governance of robotic process automation at Nordea, a large banking group operating primarily in the Nordic region. Nordea has deployed numerous software robots, for a wide range of business processes, from transaction-processing work and both internal and external reporting all the way to interaction with end users in handling of General Data Protection Regulation (GDPR)-related queries. The scene is set with a meeting where three people discuss the current state of robotic process automation implementation at Nordea: Group Head of Robotics Agnieszka Belowska Gosławska, Head of Robotic Process Automation Operations Piotr Stolarczyk and Acting Head of Robotics Execution Jaroslaw Motylewski. The presentation outlines several governance-related issues and decision points that must be addressed in connection with any deployment of robotic process automation at somewhat large scale within a company. The key issues are related to the software’s development and maintenance, robotic process automation governance and IT infrastructure. Students who have worked through the case should be able to (1) describe archetypal and hybrid governance modes for robotic process automation and (2) evaluate their advantages and disadvantages for solid infrastructure and effective software development and maintenance.


10.29007/bfmq ◽  
2019 ◽  
Author(s):  
Kaushik Hariharanagarajan ◽  
Leo Krüger ◽  
Koojana Kuladinithi ◽  
Andreas Timm-Giel

OMNeT++ as a discrete event simulator is widely used in combination with the INET framework to simulate communication networks. For students, the limited time frame designated for their master’s thesis is usually a challenge. In particular, because not only an implementation but a scientific contribution is expected as well. Within this paper we present an approach on how OMNeT++ can be used to solve the research question given in the context of a thesis. The exemplary application scenario under consideration here is based on Automated guided vehicles (AGV) which use a WLAN network to exchange information about new obstacles and their localization as such with a central server. As the performance indicators of communication networks are not yet considered while planning such setups, the effect of the network on the operation of the AGVs has to be evaluated. Especially in scenarios with a large number of AGVs the scalability of the WLAN is of interest. We show how the simulation model is parametrized based on a field analysis. With the aid of the experimental data, the different types of traffic are distinguished and modelled according to probability distributions. Based on the scenario under consideration and the envisioned large scale setups, a simulation model is implemented. In the evaluation we present exemplary results. We conclude with an explanation of the grading scheme of master’s theses to highlight how the scientific approach presented is taken into account here.


2009 ◽  
Vol 19 (03) ◽  
pp. 419-433 ◽  
Author(s):  
LAURENT LEFÈVRE ◽  
ANNE-CECILE ORGERIE

While an extensive set of research projects deal with the issue of power-saving for battery-based electronic devices, few have an interest in permanently-plugged Large-Scale Experimental Distributed Systems (LSEDS). However, a rapid study shows that each computer, member of a distributed system platform, consumes a substantial quantity of power, especially when those resources are idle. Today, given the number of processing resources involved in large-scale computing infrastructures, we are convinced that we can save a lot of electric power by proposing and applying "green policies". Introduced in this article, those policies propose to alternatively switch computer nodes On and Off in a clever way. Based on the analysis of some experimental large-scale system's usage, we propose a resource-reservation infrastructure which takes the energy issue into account. We validate our infrastructure on the large-scale experimental Grid'5000a platform and present the energy gains obtained.


Author(s):  
Ahmad Shukri Mohd Noor ◽  
Nur Farhah Mat Zian ◽  
Fatin Nurhanani M. Shaiful Bahri

<p>Distributed systems mainly provide access to a large amount of data and computational resources through a wide range of interfaces. Besides its dynamic nature, which means that resources may enter and leave the environment at any time, many distributed systems applications will be running in an environment where faults are more likely to occur due to their ever-increasing scales and the complexity. Due to diverse faults and failures conditions, fault tolerance has become a critical element for distributed computing in order for the system to perform its function correctly even in the present of faults. Replication techniques primarily concentrate on the two fault tolerance manners precisely masking the failures as well as reconfigure the system in response. This paper presents a brief survey on different replication techniques such as Read One Write All (ROWA), Quorum Consensus (QC), Tree Quorum (TQ) Protocol, Grid Configuration (GC) Protocol, Two-Replica Distribution Techniques (TRDT), Neighbour Replica Triangular Grid (NRTG) and Neighbour Replication Distributed Techniques (NRDT). These techniques have its own redeeming features and shortcoming which forms the subject matter of this survey.</p>


2020 ◽  
Vol 13 (3) ◽  
pp. 355-363
Author(s):  
Nan Xia

Whether it is a large-scale complex challenge or a radical change, it is calling for a more resilient and sustainable socio-technical system. Distributed system is a new trend of sustainable transition of socio-technical system. The research on related design strategies of distributed systems can help us better understand the nature of distributed systems, the role of designers, and help designers more calmly deal with future related challenges. This paper has conducted an in-depth understanding and discussion on the resilience of soico-technical systems and the relationship between distributed systems and resilience. We selected three representative cases, combined with a series of response measures taken by China in Wuhan during COVID-19 to analyze. Three types of distributed system design strategies suitable for China are identified.


Author(s):  
Hong Lin

Agent-oriented design has become one of the most active areas in the field of software engineering. The agent concept provides a focal point for accountability and responsibility for coping with the complexity of software systems both during design and execution (Yu, 2001). It is deemed that software engineering challenges in developing large-scale distributed systems can be overcome by an agent-based approach (Paquette, 2001). In this approach, a distributed system can be modeled as a set of autonomous, cooperating agents that communicate intelligently with one another, automate or semi-automate functional operations, and interact with human users at the right time with the right information.


Sign in / Sign up

Export Citation Format

Share Document