Search

K. Miyagoshi, Y. Teranishi, T. Kawakami, T. Yoshihisa and S. Shimojo, "Proposal of a Logical Sensor Architecture using WoT-Based Edge Microservices," 2020 IEEE 44th Annual Computers, Software, and Applications Conference (COMPSAC), Madrid, Spain, 2020, pp. 1223-1228, doi: 10.1109/COMPSAC48688.2020.00-89., 2020 IEEE 44th Annual Computers, Software, and Applications Conference (COMPSAC) [13-17 July 2020, Madrid, Spain], This paper proposes 'WoT-based Logical Sensor Architecture (WLSA),' a novel virtual sensor architecture on the basis of Web of Thing (WoT) that enables IoT applications to treat the sensor data and processing results in a uniform way in edge computing environments. WLSA reduces the computing and network resources required for the IoT applications by reusing the processing results in multiple data flows. This paper also proposes a data flow transforming (DFT) algorithm, in which a subset of the data flow is automatically replaced with a process that reuses the result of running processes so that the application developers do not need to conscious of sharing the processing result. We have implemented a prototype system and an object detection application based on WLSA using Node-RED data flow processing framework in an edge computing environment and evaluated its effectiveness. The evaluation results showed that WLSA reduced the execution time with less usage of computing resources. The prototype could accommodate 20 applications with less than 100 ms response time under 4% CPU usage.

High-performance computing (HPC) cluster systems sometimes adopt a two-layered file system composed of local and global file systems to achieve both capacity and performance in storage. In such a cluster system, the input data of an application needs to be staged from the global storage into the local storage, and the output data needs to be staged from the local storage out to the global storage. This staging operation must be efficiently and quickly performed to gain higher job throughput because an inefficient staging operation prevents waiting job requests from being executed. In particular, in the case of the cluster system with the oversubscribed interconnect shared by the storage and the computing nodes, the inter-node communication and this staging operation traffic collides, which may degrade the job throughput. In this research, we focus on the traffic collision of the inter-node communication and the staging traffic to improve job throughput, targeting the cluster system with the oversubscribed interconnect where these two types of traffic flow. In other words, whether the dynamic control of the traffic flow derived from the staging operation leads to the improvement in the job throughput or not is investigated. For the investigation, we present a traffic collision avoidance method to dynamically configure a set of data paths for each type of the traffic only while the staging operation is conducted. The evaluation in this article shows that the proposed method avoids a traffic collision and accelerates the staging operation by 22.0% on our cluster system. Also, this evaluation indicates the overhead of the application incurred by the proposed method is negligible. Furthermore, 8.7% of the job execution time is reduced by the proposed method.

We study a new service model based on the Demand Response (DR) resource provisioning at High Performance Computing (HPC) centers. This DR-based resource provisioning model allows administrators of HPC centers to provide computing services with incentives to users to compensate for the performance loss due to power saving operations. In a power conservation mode, a job’s performance may decrease, both in terms of a job waiting time and a job execution time. With DR-based resource provisioning, the submitted jobs are divided into two categories, allowed jobs and disallowed jobs, depending on the user’s tolerance in the performance degradation. The allowed jobs, if indeed affected by the power saving operations, will receive compensation in accordance with an incentive system which determines the reward to the user. For designing an appropriate demand response model, we need to focus on the increase in the job’s execution time and the job’s waiting time, and the corresponding decrease in the power consumption. These are important factors in deriving an incentive system. Currently, no existing approaches can reliably quantify the effectiveness and the contribution of these factors in HPC job scheduling and resource provisioning. In this paper, we propose a newly developed job scheduling simulator that can evaluate DR-based resource provisioning approach under various operating conditions. We designed and implemented the job scheduling simulator for HPC demand-response resource provisioning using a general-purpose discrete-event simulator. Our experiments show that the job scheduling simulator can properly represent the demand response resource provisioning using different job scheduling scenarios.