Search

A dynamic power management (DPM) method makes power-mode-related decisions based on the information available at runtime (online) or before (offline). This paper proposes energy-aware RAID scheduling methods to reduce energy consumption for distributed storage applications using both online and offline DPM strategies. The proposed energy-aware redundant array of inexpensive disk (RAID) scheduling method differs from the existing RAIDs method in that it separately strips data and parity blocks and switches the power mode from active to idle or standby while storage applications are not performing any physical I/O operations. Traditional DPM schedulers are applied for a single logical I/O operation or multiple of the logical I/O operations (workload), whereas the proposed DPM scheduler is applied for physical I/O operations. Experimental results show that the proposed storage application reduces average energy consumption by 26–36% compared to existing storage applications. [ABSTRACT FROM AUTHOR]

Recently, emerging edge computing technology has been proposed as a new paradigm that compensates for the disadvantages of the current cloud computing. In particular, edge computing is used for service applications with low latency while using local data. For this emerging technology, a neural network approach is required to run large-scale machine learning on edge servers. In this paper, we propose a pod allocation method by adding various graphics processing unit (GPU) resources to increase the efficiency of a Kubernetes-based edge server configuration using a GPU-based embedded board and a TensorFlow-based neural network service application. As a result of experiments performed on the proposed edge server, the following are inferred: 1) The bandwidth, according to the time and data size, changes in local (20.4–42.4 Mbps) and Internet environments (6.31–25.5 Mbps) for service applications. 2) When two neural network applications are run on an edge server consisted with Xavier, TX2 and Nano, the network times of the object detection application are from 112.2 ms (Xavier) to 515.8 ms (Nano); the network times of the driver profiling application are from 321.8 ms (Xavier) to 495.7 ms (Nano). 3) The proposed pod allocation method demonstrates better performance than the default pod allocation method. We observe that the number of allocatable pods on three worker nodes increases from five to seven, and compared to other papers, the proposed offloading shows similar or better response times in environments where multiple deep learning applications are implemented. [ABSTRACT FROM AUTHOR]

This paper proposes new GPU-accelerated high-performance encoding and decoding for hierarchical RAID in a multiple virtual machine environment. Pass-through GPU technology is used to provide dedicated access to GPU cores for each virtual machine, and for a virtual desktop, it also enables higher encoding and decoding performance than traditional vGPU technology. The proposed hierarchical RAID reduces the GPU overhead and resists node failure. Experimental results show that the average encoding performance of the proposed hierarchical RAID 55 improves by 11.03%, compared to another hierarchical RAID 51, with respect to various file sizes. In addition, the average disk-based decoding performance of the proposed hierarchical RAID 55 also improves by 59.61%. [ABSTRACT FROM AUTHOR]