Research Challenges in Modeling and Simulation for Engineering Complex Systems by Richard Fujimoto Conrad Bock Wei Chen Ernest Page & Jitesh H. Panchal

Author:Richard Fujimoto, Conrad Bock, Wei Chen, Ernest Page & Jitesh H. Panchal
Language: eng
Format: epub
Publisher: Springer International Publishing, Cham

4.1.4 Modeling and Simulation in the Cloud

Cloud computing offers a means to make modeling and simulation tools much more broadly accessible than was possible previously. Cloud computing provides the ability to offer modeling and simulation tools as a service that can be readily accessed by anyone with an Internet connection. In principle, users of such tools need not own their own computers and storage to complete the simulations. This feature can be especially beneficial for simulation computations requiring high-performance computing facilities because the cloud eliminates the need for simulation users to manage and maintain specialized computing equipment, a serious impediment limiting widespread adoption in the past. The “pay-as-you-go” economic model for the cloud is attractive when computational needs are heavy during certain periods of time, but much less during others. However, there are certain challenges that must be overcome for the modeling and simulation community to maximally exploit cloud computing capabilities.

Virtualization technology is used extensively in cloud computing environments. Virtualization enables one to create a “private” computational environment where resources such as CPU, memory, and operating system services appear to be readily available to applications as virtualized components . Virtualization provides isolation between applications, thereby enabling physical computing facilities to be shared among many users without concern for programs interfering with each other.

Cloud computing presents certain technical challenges, especially for parallel and distributed simulations. A significant issue that has impeded greater exploitation of public cloud computing services concerns communication delay. Both latency and latency variance, i.e., jitter, may be high in cloud computing environments and significantly degrade performance. This problem could be alleviated by improved support from cloud providers for high-performance computing . Another approach is to design parallel and distributed simulations with better ability to tolerate latency and jitter in the underlying communication infrastructure.

A second issue concerns contention for shared resources in cloud computing environments, as users are typically not guaranteed exclusive access to the computing resources used by their programs. This can lead to inefficient execution of parallel and distributed simulation codes. An approach to addressing this problem is to develop mechanisms to make these codes more resilient to changes in the underlying computing environment during the execution of the simulation. For example, dynamic load adaption is one method that can be applied to address this issue.

Cloud computing introduces issues concerning privacy and security. These are issues that are well known in the general computing community and are equally important if cloud computing is to be successfully exploited by the modeling and simulation community.

There is a trend to recognize that groups of software services require different facilities and support from cloud computing, virtualization , and service-oriented architectures. Arguably this is also true in this area and is emerging as “Modelling and Simulation as a Service” (MSaaS). This could cover modeling and simulation applications ranging from “online” simulation, where multiple users can access the same simulation (and potentially share information among them), to simulations requiring various high-performance computing support, to groups of interoperable simulations to pipelines of simulations and supporting services (real-time data collection, simulation analytics , optimizers, etc.

Download

Copyright Disclaimer:
This site does not store any files on its server. We only index and link to content provided by other sites. Please contact the content providers to delete copyright contents if any and email us, we'll remove relevant links or contents immediately.