Explore the vast range of titles available.

Power consumption is a primary concern in modern servers and data centers. Due to varying in workload types and intensities, different servers may have a different energy efficiency (EE) and energy proportionality (EP) even while having the same hardware configuration (i.e., central processing unit (CPU) generation and memory installation). For example, CPU frequency scaling and memory modules voltage scaling can significantly affect the server’s energy efficiency. In conventional virtualized data centers, the virtual machine (VM) scheduler packs VMs to servers until they saturate, without considering their energy efficiency and EP differences. In this paper we propose EASE, the Energy efficiency and proportionality Aware VM SchEduling framework containing data collection and scheduling algorithms. In the EASE framework, each server’s energy efficiency and EP characteristics are first identified by executing customized computing intensive, memory intensive, and hybrid benchmarks. Servers will be labelled and categorized with their affinity for different incoming requests according to their EP and EE characteristics. Then for each VM, EASE will undergo workload characterization procedure by tracing and monitoring their resource usage including CPU, memory, disk, and network and determine whether it is computing intensive, memory intensive, or a hybrid workload. Finally, EASE schedules VMs to servers by matching the VM’s workload type and the server’s EP and EE preference. The rationale of EASE is to schedule VMs to servers to keep them working around their peak energy efficiency point, i.e., the near optimal working range. When workload fluctuates, EASE re-schedules or migrates VMs to other servers to make sure that all the servers are running as near their optimal working range as they possibly can. The experimental results on real clusters show that EASE can save servers’ power consumption as much as 37.07%–49.98% in both homogeneous and heterogeneous clusters, while the average completion time of the computing intensive VMs increases only 0.31%–8.49%. In the heterogeneous nodes, the power consumption of the computing intensive VMs can be reduced by 44.22%. The job completion time can be saved by 53.80%.

Land-use classification is fundamental for environmental and water resource evaluation in coastal plain areas. However, comprehensive remote sensing image-based land-use analysis is challenged by the lack of massive remote sensing images and the massive computing power of large-scale server systems. In this paper, the spatial-temporal land-use change characteristics of the Hangzhou Bay area coastal plain are investigated on the Google Earth Engine platform. The proposed model uses a random forest algorithm to assist the land-use classification. The dataset is selected from the year 2009 to 2020 and classified with an average classification accuracy of 89% and Kappa coefficient of 88%. The results show that the land use in the selected region is affected by urbanization, the balance of cultivated land occupation and compensation, construction of economic development zone, and other activities. The investigation also shows that in the past 12 years, land use has changed rapidly, and each land-use type maintains the dynamic balance of occupation and compensation. Although the overall land-use distribution is stable, the information entropy fluctuates at a high level, with an average value of 1.15, and the multi-year average value of equilibrium is as high as 0.83. The driving force of land-use change is analyzed and accounted as demographics and human population dynamics, social-economic development, urbanization, and coupling effects of the above-mentioned factors.

In terms of power and energy consumption, DRAMs play a key role in a modern server system as well as processors. Although power-aware scheduling is based on the proportion of energy between DRAM and other components, when running memory-intensive applications, the energy consumption of the whole server system will be significantly affected by the non-energy proportion of DRAM. Furthermore, modern servers usually use NUMA architecture to replace the original SMP architecture to increase its memory bandwidth. It is of great significance to study the energy efficiency of these two different memory architectures. Therefore, in order to explore the power consumption characteristics of servers under memory-intensive workload, this paper evaluates the power consumption and performance of memory-intensive applications in different generations of real rack servers. Through analysis, we find that: (1) Workload intensity and concurrent execution threads affects server power consumption, but a fully utilized memory system may not necessarily bring good energy efficiency indicators. (2) Even if the memory system is not fully utilized, the memory capacity of each processor core has a significant impact on application performance and server power consumption. (3) When running memory-intensive applications, memory utilization is not always a good indicator of server power consumption. (4) The reasonable use of the NUMA architecture will improve the memory energy efficiency significantly. The experimental results show that reasonable use of NUMA architecture can improve memory efficiency by 16% compared with SMP architecture, while unreasonable use of NUMA architecture reduces memory efficiency by 13%. The findings we present in this paper provide useful insights and guidance for system designers and data center operators to help them in energy-efficiency-aware job scheduling and energy conservation.

Cloud service providers generally co-locate online services and batch jobs onto the same computer cluster, where the resources can be pooled in order to maximize data center resource utilization. Due to resource competition between batch jobs and online services, co-location frequently impairs the performance of online services. This study presents a quality of service (QoS) prediction-based scheduling model (QPSM) for co-located workloads. The performance prediction of QPSM consists of two parts: the prediction of an online service’s QoS anomaly based on XGBoost and the prediction of the completion time of an offline batch job based on random forest. On-line service QoS anomaly prediction is used to evaluate the influence of batch job mix on on-line service performance, and batch job completion time prediction is utilized to reduce the total waiting time of batch jobs. When the same number of batch jobs are scheduled in experiments using typical test sets such as CloudSuite, the scheduling time required by QPSM is reduced by about 6 h on average compared with the first-come, first-served strategy and by about 11 h compared with the random scheduling strategy. Compared with the non-co-located situation, QPSM can improve CPU resource utilization by 12.15% and memory resource utilization by 5.7% on average. Experiments show that the QPSM scheduling strategy proposed in this study can effectively guarantee the quality of online services and further improve cluster resource utilization.

With the rapid development of mobile Internet, spatial crowdsourcing has become more and more popular. Spatial crowdsourcing consists of many different types of applications, such as spatial crowd-sensing services. In terms of spatial crowd-sensing, it collects and analyzes traffic sensing data from clients like vehicles and traffic lights to construct intelligent traffic prediction models. Besides collecting sensing data, spatial crowdsourcing also includes spatial delivery services like DiDi and Uber. Appropriate task assignment and worker selection dominate the service quality for spatial crowdsourcing applications. Previous research conducted task assignments via traditional matching approaches or using simple network models. However, advanced mining methods are lacking to explore the relationship between workers, task publishers, and the spatio-temporal attributes in tasks. Therefore, in this paper, we propose a Deep Double Dueling Spatial-temporal Q Network (D3SQN) to adaptively learn the spatial-temporal relationship between task, task publishers, and workers in a dynamic environment to achieve optimal allocation. Specifically, D3SQN is revised through reinforcement learning by adding a spatial-temporal transformer that can estimate the expected state values and action advantages so as to improve the accuracy of task assignments. Extensive experiments are conducted over real data collected from DiDi and ELM, and the simulation results verify the effectiveness of our proposed models.

The increasing demand for big data processing leads to commercial off-the-shelf (COTS) and cloud-based big data analytics services. Giant cloud service vendors provide customized big data processing systems (BDPS), which are more cost-effective for operation and maintenance than self-owned platforms. End users can rent big data analytics services with a pay-as-you-go cost model. However, when users’ data size increases, they need to scale their rental BDPS in order to achieve approximately the same performance, such as task completion time and normalized system throughput. Unfortunately, there is no effective way to help end-users to choose between scale-up direction and scale-out direction to expand their existing rental BDPS. Moreover, there is no any metric to measure the scalability of BDPS, either. Furthermore, the performance of BDPS services at different time slots is not consistent due to co-location and workload placement policies in modern internet data centers. To this end, this paper proposes scalability metric for BDPS in clouds, which can mitigate the aforementioned issues. This scalability metric quantifies the scalability of BDPS consistently under different system expansion configurations. This paper also conducts experiments on real BDPS platforms and derives optimization approaches for better scalability of BDPS, such as file compression during Shuffle process in MapReduce. The experiment results demonstrate the validity of the proposed optimization strategies.

With the increasing scale of data sets and neural network models, distributed training of deep neural networks has become a trend. The main distributed parallel technology is based on expert experience, it is low efficient and hard to optimize as it needs lots of domain knowledge. There are some researchers have proposed auto-parallel technology to implement model distributed training which focused on specific models and parallel optimization factors. These methods have the problems of single factor of performance optimization, complex and low efficiency, etc. In this paper, we propose an adaptive distributed parallel training method (MP-DPS), based on the node merging of heterogeneous computing power-aware and path prediction, to search optimal parallel strategy automatically in large-scale networks. Firstly, we construct a multidimensional performance cost model to guide the design and implementation of the distributed parallel strategy. Secondly, we propose a node merging method with heterogeneous computing power awareness to reduce the search space and improve search efficiency. Finally, a graph search algorithm based on path prediction is proposed, it finds the optimal distributed parallel strategy by optimizing critical path execution time, which is based on predicting the optimal placement of critical operator node on the path. The experiments show that the deep learning model (such as ResNet, NasNet, etc.) can effectively be trained on 4 GPU and 8 GPU (P100) with the distributed parallel strategy searched by MP-DPS method, and the search time of optimal distributed parallel strategy can be reduced efficiently, compared with the FastT method.

The authors propose to use formatting templates and implicit formatting semantics information for automatic metadata identification and segmentation. The pure texts and their corresponding formatting information including line height, font type, and font size, are recognized in parallel to guide metadata identification. The authors use implicit formatting semantics, such as the change of formatting, formatting templates and implications, explicit formatting layouts, as well as predefined frequently occurred keywords database to increase the extraction accuracy. Unlike other OCR-based approaches, the authors use open source PDFBox package as the basic preprocessing tool to get pure texts and formatting values of the document contents. On top of PDFBox they built their own pipeline program, namely, PAXAT, to implement their approaches for metadata extraction. 10177 papers from arXiv, ACM, ACL and other publicly accessed and institution-subscribed sources are tested. The overall extraction accuracy of title, authors, affiliations, author-affiliation matching are 0.9798, 0.9425, 0.9298, and 0.9109, respectively.

A physiologically based pharmacokinetic (PBPK) model was developed to evaluate and predict (1) the effect of concomitant cytochrome P450 3A (CYP3A) inhibitors or inducers on the exposures of zanubrutinib, (2) the effect of zanubrutinib on the exposures of CYP3A4, CYP2C8, and CYP2B6 substrates, and (3) the impact of gastric pH changes on the pharmacokinetics of zanubrutinib. The model was developed based on physicochemical and in vitro parameters, as well as clinical data, including pharmacokinetic data in patients with B‐cell malignancies and in healthy volunteers from two clinical drug‐drug interaction (DDI) studies of zanubrutinib as a victim of CYP modulators (itraconazole, rifampicin) or a perpetrator (midazolam). This PBPK model was successfully validated to describe the observed plasma concentrations and clinical DDIs of zanubrutinib. Model predictions were generally within 1.5‐fold of the observed clinical data. The PBPK model was used to predict untested clinical scenarios; these simulations indicated that strong, moderate, and mild CYP3A inhibitors may increase zanubrutinib exposures by approximately four‐fold, two‐ to three‐fold, and <1.5‐fold, respectively. Strong and moderate CYP3A inducers may decrease zanubrutinib exposures by two‐ to three‐fold or greater. The PBPK simulations showed that clinically relevant concentrations of zanubrutinib, as a DDI perpetrator, would have no or limited impact on the enzyme activity of CYP2B6 and CYP2C8. Simulations indicated that zanubrutinib exposures are not impacted by acid‐reducing agents. Development of a PBPK model for zanubrutinib as a DDI victim and perpetrator in parallel can increase confidence in PBPK models supporting zanubrutinib label dose recommendations.

Mitochondria are the powerhouses of eukaryotic cells and the main source of reactive oxygen species (ROS) in hypoxic cells, participating in regulating redox homeostasis. The mechanism of tumor hypoxia tolerance, especially the role of mitochondria in tumor hypoxia resistance remains largely unknown. This study aimed to explore the role of mitochondria in tumor hypoxia resistance. We observed that glycolysis in hypoxic cancer cells was up-regulated more rapidly, with far lesser attenuation in aerobic oxidation, thus contributing to a more stable ATP/ADP ratio. In hypoxia, cancer cells rapidly convert hypoxia-induced O 2 · − into H 2 O 2 . H 2 O 2 is further decomposed by a relatively stronger antioxidant system, causing ROS levels to increase lesser compared to normal cells. The moderate ROS leads to an appropriate degree of autophagy, eliminating the damaged mitochondria and offering nutrients to promote mitochondria fusion, thus protects mitochondria and improves hypoxia tolerance in cancer. The functional mitochondria could enable tumor cells to flexibly switch between glycolysis and oxidative phosphorylation to meet the different physiological requirements during the hypoxia/re-oxygenation cycling of tumor growth.