Server Workloads Research Articles

Power-optimized Deployment of Key-value Stores Using Storage Class Memory

High-performance flash-based key-value stores in data-centers utilize large amounts of DRAM to cache hot data. However, motivated by the high cost and power consumption of DRAM, server designs with lower DRAM-per-compute ratio are becoming popular. These low-cost servers enable scale-out services by reducing server workload densities. This results in improvements to overall service reliability, leading to a decrease in the total cost of ownership (TCO) for scalable workloads. Nevertheless, for key-value stores with large memory footprints, these reduced DRAM servers degrade performance due to an increase in both IO utilization and data access latency. In this scenario, a standard practice to improve performance for sharded databases is to reduce the number of shards per machine, which degrades the TCO benefits of reduced DRAM low-cost servers. In this work, we explore a practical solution to improve performance and reduce the costs and power consumption of key-value stores running on DRAM-constrained servers by using Storage Class Memories (SCM). SCMs in a DIMM form factor, although slower than DRAM, are sufficiently faster than flash when serving as a large extension to DRAM. With new technologies like Compute Express Link, we can expand the memory capacity of servers with high bandwidth and low latency connectivity with SCM. In this article, we use Intel Optane PMem 100 Series SCMs (DCPMM) in AppDirect mode to extend the available memory of our existing single-socket platform deployment of RocksDB (one of the largest key-value stores at Meta). We first designed a hybrid cache in RocksDB to harness both DRAM and SCM hierarchically. We then characterized the performance of the hybrid cache for three of the largest RocksDB use cases at Meta (ChatApp, BLOB Metadata, and Hive Cache). Our results demonstrate that we can achieve up to 80% improvement in throughput and 20% improvement in P95 latency over the existing small DRAM single-socket platform, while maintaining a 43–48% cost improvement over our large DRAM dual-socket platform. To the best of our knowledge, this is the first study of the DCPMM platform in a commercial data center.

Preparedness for mass gatherings: rescue and emergency medical services\u2019 workloads during mass gathering events

BackgroundMass gathering (MG) events may cause delayed emergency responses via various mechanisms and strain the resources of local emergency services. Therefore, preparedness, including adequate pre-planning and sufficient resourcing during MG events, is vital. The aim of this retrospective register study was to investigate the impact of MG events on the workload of rescue and emergency medical service (EMS) personnel during events to enable more precise and sufficient deployment of these authorities’ operative resources.MethodsThe data from Finland covered of 25,124 EMS and rescue service missions during a three-year period (2015–2017), including data from nine MG events and reference material for the same weekdays two weeks before and after the event. The data were analysed through statistical and geospatial analyses.ResultsOur findings showed that missions increased in most events included in this study. Analysis of the missions’ reasons showed that the categories of violence, traffic accidents and other accidents and injuries increased during events, with violence-related missions showing the highest relative risk (RR 1.87, 95% CI 1.43–2.44). In the four-grade (A–D) urgency grading, the analysis showed an increase in category C missions and a decrease in non-urgent category D missions. The analysis indicated an increase in missions during the evening and night-time. The geospatial analysis revealed dense hotspots of missions in the vicinity of the event area.ConclusionThe workload for EMS and rescue service personnel increases during MG events. Most of the increase is allocated to EMS staff, peaking in evening and night hours. The geospatial analysis showed hotspots of missions on the outskirts of the actual event area during events; thus, the workload can also increase for those authority resources that are not directly allocated to the event. Detailed information regarding workloads is valuable for the authorities that are responsible for resource planning and preparedness for MG events. Replicating the study internationally would improve the methodology for the future.

The Effect of Physician Communication on Inpatient Satisfaction.

(1) Background: The importance of physician-patient communication and its effect on patient satisfaction has become a hot topic and has been studied from various aspects in recent years. However, there is a lack of systematic reviews to integrate recent research findings into patient satisfaction studies with physician communication. Therefore, this study aims to systematically examine physician communication’s effect on patient satisfaction in public hospitals. (2) Methods: Using a keywords search, data was collected from five databases for the papers published until October 2021. Original studies, observational studies, intervention studies, cross-sectional studies, cohort studies, experimental studies, and qualitative studies published in English, peer-reviewed research, and inpatients who communicated with the physician in a hospital met the inclusion criteria. (3) Results: Overall, 11 studies met the inclusion criteria from the 4810 articles found in the database. Physicians and organizations can influence two determinants of inpatient satisfaction in physician communication. Determinants of patient satisfaction that physicians influence consist of amounts of time spent with the patient, verbal and nonverbal indirect interpersonal communication, and understanding the demands of patients. The organization can improve patient satisfaction with physician communication by the organization’s availability of interpreter service and physician workload. Physicians’ communication with inpatients can affect patient satisfaction with hospital services. (4) Conclusions: To improve patient satisfaction with physician communication, physicians and organizational determinants must be considered.

Distributed deep learning training using silicon photonic switched architectures

The scaling trends of deep learning models and distributed training workloads are challenging network capacities in today’s datacenters and high-performance computing (HPC) systems. We propose a system architecture that leverages silicon photonic (SiP) switch-enabled server regrouping using bandwidth steering to tackle the challenges and accelerate distributed deep learning training. In addition, our proposed system architecture utilizes a highly integrated operating system-based SiP switch control scheme to reduce implementation complexity. To demonstrate the feasibility of our proposal, we built an experimental testbed with a SiP switch-enabled reconfigurable fat tree topology and evaluated the network performance of distributed ring all-reduce and parameter server workloads. The experimental results show up to 3.6× improvements over the static non-reconfigurable fat tree. Our large-scale simulation results show that server regrouping can deliver up to 2.3× flow throughput improvement for a 2× tapered fat tree and a further 11% improvement when higher-layer bandwidth steering is employed. The collective results show the potential of integrating SiP switches into datacenters and HPC systems to accelerate distributed deep learning training.

A Reliable Load Balancing Fault Tolerant Multi-SDN Controller approach in a typical Software Defined Network

INTRODUCTION: In the 21st Century, we have an emerging, modernized networking approach for efficient transfer of data in the form of packets i.e Software defined Networking (SDN). It is often recognized as the network paradigm acquired by an isolated independent control plane by the means of the underlying networking interface. As the communication medium is getting rapidly increased, the usage of the number of resources are also increasing in the same ratio, this results to connectivity problem in communication networks as well. If the development in technology is gaining more interest, the resource utilization will also increase.OBJECTIVES: In the proposed work here, we need to establish the connection between the client (or the end user) and the server by calculating the network performance using the parameters such as server response time and the throughput of the network by the method of load balancing.METHODS: This paper aims to identify the new methods or techniques to improve the performance of SDN with the increase in reliability, specifically required to improve the real-time data delivery. Alternatively, a reliable disjoint, distributed Multi- SDN controller will be used, to collect the updated network information like packet size, connectivity, packet flow etc. The methods involved are ideal controller optimization technique, distruption acquiscent technique and Maximum flow shortest route technique.RESULTS: From the results obtained by the simulation, the Reliable Load balancing Fault Tolerant Multi-SDN Controller(RLFTMS) approach is more reliable, fault tolerant compared to the other existing traditional mechanisms.CONCLUSION: Load balancing based on the server response time, throughput is being compared , this improves the server workload ditribution among various datacenters. Ultimately the customers or clients will have a great experience in terms of the reducing response time and increase in the throughput.

A Profit-Aware Coalition Game for Cooperative Content Caching at the Network Edge

The user demands to delay-sensitive applications have put forward new requirements for the mobile networks. Edge caching as a promising way is proposed to enhance the Quality of Service (QoS) for end users at the network edge. Given the widely distributed edge nodes, the content providers (CPs) usually prefer to integrate them with cooperative caching services, by forming a cache coalition. Although such a coalition could be beneficial as a whole, it neglects the profits of individual members, which is one major concern in forming the coalition itself. Besides, due to the poor scalability, the conventional cooperation scheme, which only considers fixed edge nodes, cannot adapt to the spatial and temporal imbalance of user requests. In this article, we tackle the problems of coalition establishment and profit allocation among the coalition members. Particularly, by adopting both fixed edge nodes and mobile vehicles as caching nodes, we propose a hybrid service provisioning framework and cooperative service caching and workload scheduling methods. To maximize the profits in managing the caching resources in the established coalition, we devise an optimization model with a mixed-integer programming (MIP), in which the QoS requirements of end users and caching capacities of each coalition member are also considered as constraints. In addition, based on the Hedonic game theory, we propose a dynamic coalition algorithm to guide each member to join or leave the coalition at each time slot out of its own profits. The experimental results demonstrate that compared to the cases only considering fixed caching nodes, our hybrid caching scheme can improve: 1) the overall profit of the coalition by 53% and 2) the average profit of individual participants by 42%, respectively.

Two-Stage Robust Edge Service Placement and Sizing Under Demand Uncertainty

Edge computing has emerged as a key technology to reduce network traffic, improve user experience, and enable numerous Internet of Things applications. In this article, we study an optimal resource procurement problem for a service provider (SP), who can purchase resources from various edge nodes in the edge computing market to serve its users’ requests. How to jointly optimize the service placement, resource sizing, and workload allocation decisions is a challenging problem, which becomes even more complicated when considering demand uncertainty. To this end, we propose a novel two-stage adaptive robust optimization framework to help the SP optimally determine the locations for installing its service (i.e., placement) and the amount of computing resource to purchase from each location (i.e., sizing). The proposed placement and sizing solution can hedge against any possible realization within a predefined demand uncertainty set. Given the first-stage robust solution, the optimal resource and workload allocation decisions are computed in the second stage after the uncertainty is revealed. To solve the two-stage model, this article presents an iterative solution approach by employing the column-and-constraint generation method that decomposes the underlying problem into a master problem and a max–min subproblem associated with the second stage. Extensive numerical results are shown to illustrate the efficacy of the proposed model.

Service Quality and Efficiency-Oriented Training Design in Air Traffic Management

The main objective of the air traffic control (ATC) services is to ensure safe, efficient, and expeditious air traffic flow within an airspace or aerodrome according to international rules and regulations. Despite, the development of various automation systems to keep up with these objectives, air traffic control services still rely on radio-telephony (RT) communication which requires exchange of instructions and readbacks between air traffic controllers (ATCOs) and pilots. Although various studies indicate that the less number of instructions are correlated with less a workload and better quality of air traffic services, this is no always so, especially for the high traffic levels at which more instructions are required to maintain safe separation and efficient air traffic flow. Number of instructions can also differentiate depending on the characteristics of airspace and the type of air traffic control (area, approach or aerodrome). In order to obtain a broader insight regarding the quality of service including the air traffic and airspace characteristics, we need to include the effects of flight efficiency in the analysis. This study aims to explore correlation between the quality of service and flight efficiency based on real-time air traffic simulation experiments done on ATCO trainees in ATCOSIMA project. The experiments were carried out on two groups of trainees. The first group had only formal ATC training background while the second group went under an additional training focused on the flight efficiency and the service quality in addition to their formal training background. During the experiments, number of instructions per flight was collected as the indicator of quality of service while the total completion time of flights were collected as the indicator of the flight efficiency. The results of the analysis indicated that the number of instructions was decreased with the additional training but no significant change was observed in the flight completion time. A moderately significant correlation was also observed between the average number of instructions per aircraft and the completion time. In conclusion, a new training method focused on flight efficiency in addition to the formal training improve the quality of service and efficiency level of trainees. The findings of the study can provide important insights for the development of ATCO training and assessment. For the future work, the analysis can be expended for more efficiency indicators including effectiveness of trajectories and environmental impacts.

Fall Detection System With Artificial Intelligence-Based Edge Computing

Falls are the second leading cause of death from unintentional injuries in older adults. Although many systems have been used to detect falls, they are limited by the computational complexity of the algorithm. The images taken by the camera must be transmitted through a network to the back-end server for calculation. As the demand for Internet of Things increases, this architecture faces such problems as high bandwidth costs and server computing overload. Emerging methods reduce the workload of servers by transferring certain computing tasks from cloud servers to edge computing platforms. To this end, this study developed a fall detection system based on neuromorphic computing hardware, which streamlines and transplants the neural network model of the back-end computer to the edge computing platform. Through the neural network model with integer 8 bit precision deployed on the edge computing platform, the object photos obtained by the camera are converted into human motion features, and a support vector machine is then used for classification. After experimental evaluation, an accuracy of 96% was reached, the detection speed of the overall system was 11.5 frames per second, and the power consumption was 0.3 W. This system can monitor the fall events of older adults in real time and over a long period. All data were calculated on the edge computing platform. The system only reports fall events via Wi-Fi, thereby protecting the privacy of the user.

Vascular Access Teams: A Global Outlook on Challenges, Benefits, Opportunities, and Future Perspectives

Highlights Abstract Background: Specialized vascular access training for medical professionals organized into vascular access teams (VATs) was shown to improve patient outcomes, clinical efficiency, and cost savings. Professional perspectives on VAT benefits, organization, challenges, and opportunities on a global scale remain inadequately explored. Using detailed perspectives, in this study, we explored the global VAT landscape, including challenges faced, clinical and clinico-economic impacts of VATs, with emphasis on underresearched facets of VAT initiation, data dissemination, and metrics or benchmarks for VAT success. Methods: Semistructured in-depth interviews of 14 VAT professionals from 9 countries and 5 continents were used to elicit qualitative and quantitative information. Results: Catheter insertions (100%) and training (86%) were the most performed VAT functions. Based on a 1–7 scale evaluating observed impacts of VATs, patient satisfaction (6.5) and institutional costs (6.2) were ranked the highest. VAT co-initiatives, advanced technology utilization (6.6), and ongoing member training (6.3) distinctly impacted VAT endeavors. Most institutions (64%) did not have routine mechanisms for recording VAT-related data; however, all participants (100%) stated the importance of sharing data to demonstrate VAT impacts. Time constraints (57%) emerged as one of the major deterrents to data collection or dissemination. The majority (64%) experienced an increased demand or workload for VAT services during the COVID-19 pandemic. Conclusions: Despite the global variances in VATs and gaps in VAT-related data, all participants unanimously endorsed the benefits of VAT programs. Evaluating the impact of VATs, disseminating VAT-related data, and forging specialized institutional partnerships for data sharing and training are potential strategies to tackle the hurdles surrounding VAT formation and sustenance.

A data-assisted first-principle approach to modeling server outlet temperature in air free-cooled data centers

The server outlet temperature is an important thermal condition to the operation of an air free-cooled data center that uses fans to continuously pass the outside air through the server room to cool the computing devices. However, the standard server’s management and monitoring tool cannot read the server’s built-in outlet temperature sensors fast enough to catch up the fast dynamics of the server outlet thermal condition caused by the changing server workload. Moreover, many server models do not have built-in sensors that can measure the server outlet temperature. In this paper, we develop a data-assisted first-principle model that leverages available built-in sensors and server’s operating monitoring tools to achieve low-latency estimation of the server outlet temperature. Specifically, the developed model takes the inlet and processor core temperatures, server’s fan speed, and processor utilization which are measured by hardware/software sensors as inputs to predict the outlet temperature with low latencies. Our extensive evaluation based on real data traces collected from a real air free-cooled data center testbed shows that our model can accurately predict the outlet temperature with an average root mean squared error ranging from 1.21 °C to 1.46 °C under various cold supply air temperatures and processor utilization levels.

Assessment and management of issues in early pregnancy

Approximately 5% of the workload of UK emergency ambulance services involves managing obstetric patients. This places pregnancy firmly within the scope of prehospital care but training often focuses on critical illness during pregnancy rather than the range of presentations seen. This clinical review aims to discuss the implications of early pregnancy with a focus on ectopic pregnancy, rhesus incompatibility, miscarriage and hyperemesis gravidarum. Normal presentations of pregnancy and alternative management strategies, including early pregnancy units, are also discussed.

Deep Reinforcement Learning-Based Server Selection for Mobile Edge Computing

With Mobile Edge Computing (MEC), computational intensive applications can be offloaded to the nearby edge servers to support latency-sensitive applications on mobile devices. Different from the cloud, edge servers usually have limited resources, and then selecting which edge server to run the offloaded computation becomes an important issue. Although server selection has received considerable attention, not much work has been done to consider the limited coverage of the edge server and the frequent user movement, which introduce many dynamic changing factors affecting the workload of the edge server and making it hard to achieve long-term optimum in the edge server selection. To deal with these challenges, we model the problem of continuous server selection as a Markov Decision Process (MDP). The difficulty of this problem is that achieving long-term optimum requires future knowledge, such as user mobility, server workload, etc, which is not known a priori. We do not have such knowledge and thus cannot find the optimal policy through traditional methods. To address this problem, we propose a <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Deep Reinforcement Learning (DRL)</i> based algorithm to learn the selection policy based on the observed performance of past server selections. Specifically, a Long Short-Term Memory (LSTM) based neural network is exploited to encode the historical information which helps infer future knowledge of the dynamically changing factors. Then the DRL model selects the optimal server automatically based on the extracted system states. Extensive trace-driven evaluations demonstrate that the proposed DRL-based algorithm has the lowest overall cost compared to existing solutions.

Health-related quality of life among intensive care unit workers

ObjectiveTo analyse health-related quality of life in a sample of doctors and nurses from 2 intensive care units in the city of Buenos Aires, Argentina. MethodsAn analytical, cross-sectional and quantitative study. The Spanish version of the SF-36 questionnaire was used, which comprises 8 dimensions grouped into 2 components (physical and mental) and evaluates the health status of respondents on a scale from 0 to 100; the higher the score, the better the perception of health status. ResultsNinety-five professionals from 2 intensive care units participated, 58% from a private institution and the rest from a public institution. The majority were women (57%), doctors (37%) with an average age of 42 years and with a maximum undergraduate degree and degree (59%). Of the dimensions making up health-related quality of life, the vitality dimension was perceived poorest with an average of 53.3, and physical function was the best evaluated at 87.8. The physical component dimensions were better evaluated than those of the mental component. The variables sex (male), age (> 40 years), presence of chronic diseases, qualification (medical), hourly workload (> 40 h per week) and years of service (> 11 years), seem to have a negative influence on the perception of health-related quality of life. ConclusionsStrategies need to be implemented to reduce the impact of work activity on the mental and physical health of health professionals. Strategies for health promotion and the prevention of known psychosocial and ergonomic risks are required, and workloads and duration of working hours need to be adjusted to preserve the integrity of health workers.

Process, process, process: the OCR watchwords

As the primary enforcement authority for Section 504 of the Rehabilitation Act, the Office for Civil Rights routinely reviews complaints by postsecondary students who have requested but have been denied accommodations, academic adjustments, and auxiliary aids. Many accommodation requests that make up the workload of disability services offices center around test accommodations, attendance and assignment accommodations, and housing accommodations, rather than auxiliary aids.

Analysis And Evaluation Of Pharmacist Power Needs With The WISN Method In The Pharmaceutical Installation Of Haji Hospital Surabaya

The success of the hospital in carrying out its functions is influenced by several factors, among which the most dominant is human resources. Human resources are one of the important factors to be able to provide quality and comprehensive services in health services and are the key determinants of the rate and quality of services. The method of calculating needs based on workload (WISN) is an indicator that shows the amount of workforce required for health facilities based on workload, so that the allocation / relocation of labor will be easier and more rational. This research is a descriptive observational research with qualitative data analysis with purposive sampling method, namely the sampling technique by determining certain criteria. The HR category that will be calculated is the pharmacist in the Pharmacy Installation of the Surabaya Haji Hospital and results in a calculation of the total need for pharmacists of 24 people, compared to the current number of pharmacists of 20 pharmacists, the addition of pharmacists is 4 people in the Home Pharmacy Installation Haji Hospital Surabaya. The current number of fulfilled personnel from the calculation results only fulfills 83% of the ideal amount generated by the WISN method. With the obstacles faced such as the large workload of clinical pharmacy services that are carried out every day. To anticipate this, it is necessary to fulfill the need for pharmacists.

Study of the Workload and Human Resources Demand of Basic Public Health Services in Township Hospitals of Sichuan Province

To understand the basic public health service (BPHS) tasks and workload of the township hospitals in Sichuan Province and to provide empirical evidence for improving the allocation of human resources (HR) and the quality of BPHS in township hospitals. Retrospective survey was conducted to collect information regarding the actual working hours per unit time of service of BPHS. Information on the service volume and unit service time of BPHS tasks for 2018 was collected in 7 township hospitals in an economically developed area and an area of moderate economic development. Estimated HR required to cover the BPHS tasks was calculated based on the survey data of actual work time. The time required for the basic units of various BPHS tasks varied greatly. The top four most time-consuming services showing consistent results in the the two regions were health education, health supervision and coordination, management of children aged 0-6 and maternal care management. Institutions in the area of moderate economic development reported higher per capita service volume in the registration of children for vaccination, management of hypertension, and health supervision and coordination than those in the economically developed area, but lower service volume for other services. There is a shortage of full-time staffs in both areas. Differences in the service volume of BPHS exist in regions of different levels of economic development. The shortage in HR could be related to the lower quality and volume of the service. Comprehensive exploration of policies to recruit and retain competent personnel, the strengthening of grassroots professional training, and optimization of information technology will help improve the competence and quality of grassroots services.

Optimal cloud assistance policy of end-edge-cloud ecosystem for mitigating edge distributed denial of service attacks

Edge computing has become a fundamental technology for Internet of Things (IoT) applications. To provide reliable services for latency-sensitive applications, edge servers must respond to end devices within the shortest amount of time possible. Edge distributed denial-of-service (DDoS) attacks, which render edge servers unusable by legitimate IoT applications by sending heavy requests from distributed attacking sources, is a threat that leads to severe latency. To protect edge servers from DDoS attacks, a hybrid computing paradigm known as an end-edge-cloud ecosystem provides a possible solution. Cloud assistance is allowed with this architecture. Edge servers can upload their pending tasks onto a cloud center for a workload reduction when encountering a DDoS attack, similar to borrowing resources from the cloud. Nevertheless, before using the ecosystem to mitigate edge DDoS attacks, we must address the core problem that edge servers must decide when and to what extent they should upload tasks to the cloud center. In this study, we focus on the design of optimal cloud assistance policies. First, we propose an edge workload evolution model that describes how the workload of the edge servers change over time with a given cloud assistance policy. On this basis, we quantify the effectiveness of the policy by using the resulting overall latency and formulate an optimal control problem for seeking optimal policies that can minimize such latency. We then provide solutions by deriving the optimality system and discuss some properties of the optimal solutions to accelerate the problem solving. Next, we introduce a numerical iterative algorithm to seek solutions that can satisfy the optimality system. Finally, we provide several illustrative numerical examples. The results show that the optimal policies obtained can effectively mitigate edge DDoS attacks.

A multi-objective consistent home healthcare routing and scheduling problem in an uncertain environment

Home health care (HHC) companies serve as the alternative to hospitals aiming to provide customers with medical care at home. A crucial challenge for HHC providers is to optimize routes and schedules for their caregivers to serve customers. Inspired by the practices in the HHC industry, this paper addresses a multi-objective home healthcare routing and scheduling problem (HHRSP) with several conflicting objectives: minimizing routing cost and improving service consistency and workload balance. We refer to the problem as a multi-objective consistent home healthcare routing and scheduling problem (MoConHHRSP). To be more practical, uncertain travel and service times are also considered and defined based on uncertainty theory. Next, the uncertain programming model for the proposed MoConHHRSP is formulated and then reduced to its deterministic equivalent. Due to the NP-hard essence of the problem, an improved multi-objective artificial bee colony (IMOABC) metaheuristic, integrating the large neighborhood search heuristic and an adapted non-dominated solution set update strategy into the multi-objective artificial bee colony (MOABC) framework, is developed. Finally, a series of numerical experiments are conducted to illustrate the competitive performance of the designed algorithm by comparing it with other multi-objective algorithms from multiple evaluation metrics. Furthermore, the trade-off analysis reveals that a better caregiver consistency can be achieved at a high price of total costs and workload balance, while a great improvement on the workload balance can be provided with little deterioration in caregiver consistency. In many cases, low total costs and a high level of workload balance can be achieved simultaneously.

Process, process, process: the OCR watchwords

As the primary enforcement authority for Section 504 of the Rehabilitation Act, the Office for Civil Rights routinely reviews complaints by postsecondary students who have requested but have been denied accommodations, academic adjustments, and auxiliary aids. Many accommodation requests that make up the workload of disability services offices center around test accommodations, attendance and assignment accommodations, and housing accommodations, rather than auxiliary aids.