Theory and Practice in Parallel Job Scheduling
IPPS '97 Proceedings of the Job Scheduling Strategies for Parallel Processing
Optimistic Synchronization of Parallel Simulations in Cloud Computing Environments
CLOUD '09 Proceedings of the 2009 IEEE International Conference on Cloud Computing
Performance Analysis of Cloud Computing Services for Many-Tasks Scientific Computing
IEEE Transactions on Parallel and Distributed Systems
GCMS '09 Proceedings of the 2009 Grand Challenges in Modeling & Simulation Conference
| Hi-index | 0.00 |
The cloud computing paradigm attracts increasing amount of Modeling&Simulation (M&S) practitioners to perform their simulations in the cloud. Two issues, namely, the architecture of the Cloud-based Simulation (CSim) and the parallel simulation job scheduling in the CSim, should be addressed ï卢聛rst to make the CSim practical. This paper reports our recent work on the two issues. The architecture we proposed covers the software involved in the whole process of M&S by providing the Modeling as a Service (MaaS), the Execution as a Service (EaaS) and the Analysis as a Service (AaaS). The architecture also encourages the reuse of available simulation resources with the aid of the Simulation Resource as a Service (SRaaS). For the issue of parallel simulation job scheduling in the CSim, we ï卢聛rst propose a two-tier processor partition method to organize virtual machines (VMs) for parallel simulation workload consolidation, the two-tier VMs have different CPU priority. We then present four scheduling algorithms under such a partition method to cope with four common situations. Our extensive experiments on well-known traces show that all the four algorithms signiï卢聛cantly outperform their competitors.