Workload Fluctuations Research Articles

Dynamic Provisioning of Service Composition in a Multi-Tenant SaaS Environment

Multi-tenant service composition has become a common delivery model for business processes in cloud computing. To dynamically support the workload tenant variation, elasticity holds the promise of ensuring the quality of service (QoS) of the business process by providing the involved service instances at a low cost. However, integrating both of multi-tenancy and elasticity during service composition is a key problem for serving multiple tenants from a single process instance. Nowadays, existing approaches in the field of cloud service composition, although numerous, still fall short since they cannot adequately address issues related to supporting the scalability of the composed service and adapting it to the workload fluctuation. In this paper, we propose a holistic approach which makes the dynamic multi-tenant services matching and manages their elasticity in distributed business processes. This approach is based on a generic service pattern that integrates multi-tenancy property and handles elasticity at the process and service levels. Furthermore, we present elastic composition algorithms to compose multi-tenant cloud services and perform their elasticity through the proposed service pattern. The evaluation of our approach, compared to the baseline approach, proves that the latency taken to provide an elastic multi-tenant service composition and detect its SLA (Service Level Agreements) violation are reasonably short. We also show that the CPU overhead of using our approach is negligible. Furthermore, experimental results demonstrate the merits of our approach in terms of minimizing the memory consumption through the deployed service instances.

PRIMA: Subscriber-Driven Interference Mitigation for Cloud Services

Network services, e.g., video streaming services, are increasingly being deployed on public cloud platforms. Such services often employ horizontal scaling where a group of resource instances, e.g., virtual machines (VMs), handle incoming workload. The response time of such services is often affected by interference, i.e., contention among resource instances belonging to multiple cloud subscribers for shared cloud resources. Most commercial cloud platforms do not support built-in mechanisms to detect interference and mitigate its impact. Consequently, subscribers of such platforms, i.e., network service providers, need to deploy their own mechanisms to ensure a specified end user response time target is continuously met even in the face of fluctuations in workload and interference. This paper describes PRIMA, our implementation of such a mechanism. PRIMA uses automated and controlled performance tests to build models that capture the joint impact of workload and interference on the response time of each resource instance employed by a service. It adapts the system to changing workload and interference conditions by using these models at runtime to control the number of instances in the system and the distribution of load among these instances. Unlike existing subscriber-oriented interference mitigation techniques in literature, PRIMA guarantees that a subscriber-specified response time threshold is satisfied at every resource instance assigned to a service. Furthermore, in contrast to these approaches PRIMA can help a subscriber avoid using more instances than necessary by automatically selecting at runtime the least number of instances required for handling the observed workload and interference. We experimentally validate the effectiveness of PRIMA in both private and public cloud environments. Results show that PRIMA outperforms competing approaches proposed by us and others, including those that are commonly used in practice. They also reveal that PRIMA can automatically calibrate its models at runtime to account for any model prediction errors.

Intelligent and Agile Control of Edge Resources for Latency-Sensitive IoT Services

This paper presents an intelligent and agile resource control scheme for a latency-sensitive virtual network function (VNF) of Internet of things directory service (IoT-DS) deployed in a virtualized edge cloud whose computational and networking resources can be adjusted dynamically. The objective of the proposed scheme is to adjust resources dynamically such that the IoT-DS function can resolve IoT queries and provide IoT records within a bounded delay for latency-sensitive services such as automated driving, despite fluctuations in workloads. The proposed scheme leverages multiple regression models for resource demand prediction and dynamic adjustment. These models are trained offline before their deployment with a large training dataset collected from the system operating with simulated workloads. After the deployment, they are updated regularly by online retraining for more accurate performances. We aim to optimize resource allocation to satisfy both the target performance in terms of service latency and resource utilization. The results obtained from an experimental system implementation of the IoT-DS function in Docker containers show that the dynamic adjustment of CPU resources by the proposed scheme with supervised offline training reduces the CPU resource demand by 21.9% and the number of lookup latency requirement violations by 58.2% in comparison with a threshold rule-based conventional algorithm. Moreover, the proposed scheme can offer an agile control of CPU resources within a 1 s interval, which is five times faster than those reported in previous studies. The addition of unsupervised online retraining further reduces CPU resource requirements by 52% and lookup latency requirement violation cases by 62.5% compared with when no adjustments are performed.

Cluster-based optimal VM placement using crow search algorithm for cloud data centres

The extensive use of computational power from data centres causes huge energy consumption. A good virtual machine (VM) placement strategy would make better consolidation of VMs in a data centre (DC) that reduces energy consumption. However, it is hard to balance hosts in DCs due to the workload fluctuation by application and scaling of VMs. The optimal decision on VM placement and consolidation is an NP-hard problem and many researchers have proposed solutions to tackle this problem but they lack efficient exploitation of the mechanisms. Therefore, this paper proposes a hierarchical cluster-based approach with a meta-heuristic crow search algorithm (CSA) for the optimal selection of hosts to place the VMs and consolidate the maximum number of VMs on a minimum number of hosts. The work is simulated in CloudSim using real workload traces. Experimental results show that proposed work reduces energy consumption, SLA violations and VM migrations while ensuring better resource utilisation.

Workload forecasting based elastic resource management in edge cloud

Cloud services are provided at the edge of the network so that data from users can be processed and calculated at the edges. The user's irregular access triggers the fluctuations of the edge cloud workload. Therefore, an elastic resource management method based on workload forecasting in edge clouds is proposed in this paper. When the resource demand is large, more resources are requested from the cloud service provider so that the task can be completed before the deadline. When the resource demand is small, the idle resource is released to meet the cost constraint. The resource demand is judged based on the workload forecasting. In order to improve the accuracy of workload forecasting, a workload forecasting model based on error correction is proposed in this paper. Neither overload nor the light-load status of edge cloud nodes can make full use of the resources. To improve the node processing performance and reduce migration times, a workload migration model for minimizing migration times is proposed in this paper. The experimental results show that the proposed methods can effectively forecast the workload and improve the processing performance of the entire cluster.

Machine learning-based auto-scaling for containerized applications

Containers are shaping the new era of cloud applications due to their key benefits such as lightweight, very quick to launch, consuming minimum resources to run an application which reduces cost, and can be easily and rapidly scaled up/down as per workload requirements. However, container-based cloud applications require sophisticated auto-scaling methods that automatically and in a timely manner provision and de-provision cloud resources without human intervention in response to dynamic fluctuations in workload. To address this challenge, in this paper, we propose a proactive machine learning-based approach to perform auto-scaling of Docker containers in response to dynamic workload changes at runtime. The proposed auto-scaler architecture follows the commonly abstracted four steps: monitor, analyze, plan, and execute the control loop. The monitor component continuously collects different types of data (HTTP request statistics, CPU, and memory utilization) that are needed during the analysis and planning phase to determine proper scaling actions. We employ in analysis phase a concise yet fast, adaptive, and accurate prediction model based on long short-term memory (LSTM) neural network to predict future HTTP workload to determine the number of containers needed to handle requests ahead of time to eliminate delays caused by starting or stopping running containers. Moreover, in the planning phase, the proposed gradually decreasing strategy avoids oscillations which happens when scaling operations are too frequent. Experimental results using realistic workload show that the prediction accuracy of LSTM model is as accurate as auto-regression integrated moving average model but offers 600 times prediction speedup. Moreover, as compared with artificial neural network model, LSTM model performs better in terms of auto-scaler metrics related to provisioning and elastic speedup. In addition, it was observed that when LSTM model is used, the predicted workload helped in using the minimum number of replicas to handle future workload. In the experiments, the use of GDS showed promising results in keeping the desired performance at reduced cost to handle cases with sudden workload increase/decrease.

Modeling Pathology Workload and Complexity to Manage Risks and Improve Patient Quality and Safety

Anatomic pathology (AP) laboratories provide critical diagnostic information that help determine patient treatments and outcomes, but the risks of AP operations and their impact on patient safety and quality of care remain poorly recognized and undermanaged. Hospital-based laboratories face an operational and risk management challenge because clinical work of unknown quantity and complexity arrives with little advance notice, which results in fluctuations in workload that can push operations beyond planned capacity, leading to diagnostic delays and potential errors. Modeling the dynamics of workload and complexity in AP offers the opportunity to better use available information to manage risks. We developed a stock-and-flow model of a typical AP laboratory operation and identified key exogenous inputs that drive AP work. To test the model, we generated training and validations data sets by combining data from the electronic medical records and laboratory information systems over multiple years. We demonstrate the implementation of 10-day AP work forecast generated on a daily basis, and show its performance in comparison with actual work. Although the model somewhat underpredicts work as currently implemented, it provides a framework for prospective management of resources to ensure quality during workload surges. Although full implementation requires additional model development, we show that AP workload largely depends on few and accessible clinical inputs. Recognizing that level loading of work in a hospital is not practical, predictive modeling of work can empower laboratories to triage, schedule, or mobilize resources more effectively and better manage risks that reduce the quality or timeliness of diagnostic information.

A self‐learning fuzzy approach for proactive resource provisioning in cloud environment

SummaryThe development of a communication infrastructure has made possible the expansion of the popular massively multiplayer online games. In these games, players all over the world can interact with one another in a virtual environment. The arrival rate of new players to the game environment causes fluctuations and players always expect services to be available and offer an acceptable service‐level agreement (SLA), especially in terms of response time and cost. Cloud computing emerged in the recent years as a scalable alternative to respond to the dynamic changes of the workload. In massively multiplayer online games applications, players are allowed to lease resources from a cloud provider in an on‐demand basis model. Proactive management of cloud resources in the face of workload fluctuations and dynamism upon the arrival of players are challenging issues. This paper presents a self‐learning fuzzy approach for proactive resource provisioning in cloud environment, where key is to predict parameters of the probability distribution of the incoming players in each period. In addition, we propose a self‐learning fuzzy autoscaling decision‐maker algorithm to compute the proper number of resources to be allocated to each tier in the massively multiplayer online games by applying the predicted workload and user SLA. We evaluate the effectiveness of the proposed approach under real and synthetic workloads. The experimental results indicate that the proposed approach is able to allocate resources more efficiently than other approaches.

Monitoring Workload in Elite Female Basketball Players During the In-Season Phase: Weekly Fluctuations and Effect of Playing Time.

To assess the weekly fluctuations in workload and differences in workload according to playing time in elite female basketball. Twenty-nine female basketball players (mean ± standard deviation, age: 21±5yr; stature: 181±7cm; body mass: 71±7kg; and playing experience: 12±5yr) belonging to the 7 female basketball teams competing in the first division Lithuanian Women's Basketball League (LMKL) were recruited. Individualized training loads (TL) and game loads (GL) were assessed using the session-RPE following each training session and game during the entire in-season phase (24 weeks). Percentage (%) changes in total weekly TL (weekly TL+GL), weekly TL, weekly GL, chronic workload, acute:chronic workload ratio (ACWR), training monotony, and training strain were calculated. Mixed linear models were used to assess differences for each dependent variable, with playing time (low vs high) used as fixed factor and subject, week, and team as random factors. The highest changes in total weekly TL, weekly TL, and ACWR were evident in week 13 (47%, 120%, and 49% respectively). Chronic workload showed weekly changes ≤10%, while monotony and training strain registered highest fluctuations in weeks 17 (34%) and 15 (59%), respectively. A statistically significant difference in GL was evident between players completing low and high playing times (p=0.026, moderate), while no significant differences (p>0.05) were found for all other dependent variables. Coaches of elite female basketball teams should monitor weekly changes in workload during the in-season phase to identify weeks that may predispose players to unwanted spikes and adjust player workload according to playing time.

Pec: Proactive Elastic Collaborative Resource Scheduling in Data Stream Processing

In the Distributed Parallel Stream Processing Systems (DPSPS), elastic resource allocation allows applications to dynamically response to workload fluctuations. However, resource provisioning can be particularly challenging, due to the unpredictability of the workload. In addition, unlike CPU resources, bandwidth resources are often ignored in resource allocation. Moreover, resource allocation and resource placement are considered separately. In this paper, we investigate the proactive elastic resource scheduling problem for computation-intensive and communication-intensive applications, which aims at meeting the latency requirement with the minimal energy cost, and propose a dynamic collaborative strategy from the systemic perspective. Specifically, we first model a collaborative workload prediction pattern to accurately predict the upcoming workload, and construct a latency estimation model to estimate the latency of the application. Then, we design an energy-efficient resource pre-allocation method, in which the CPU frequency adjustment and the stability of resource reconfigurations are both considered. Finally, we present a communication-aware resource placement approach. Simulation results show that, compared with the reactive strategies, our strategy achieves an obviously better latency performance, and effectively avoids unnecessary resource adjustments. Meanwhile, the energy consumption is about saved by 50 percent on average, and the communication cost is maintained at a very low level of 4 percent.

An optimal checkpointing model with online OCI adjustment for stream processing applications

SummaryCheckpoint‐based fault‐tolerant (FT) methods have been widely used to enhance the reliability of stream processing systems, but a checkpointing process usually introduces considerable overhead. It is a critical issue to choose the optimal checkpoint interval (OCI) that maximizes the processing efficiency. Traditional OCI models consider the recovery time equals to the execution time from the last checkpoint to the failure moment. However, for stream processing jobs, the recovery time is related to reprocessing workloads, depending on the real‐time input data before a failure. A new model is needed to choose the OCI for stream processing applications. Moreover, the input data rate of a stream processing job fluctuates over time. To solve these problems, we present a novel DSPS OCI (DOCI) model in this paper. We prove that it maximizes the processing efficiency for a given time. We propose an approach to dynamically adjust the OCI for an application to accommodate the workload fluctuations. We conduct simulation experiments to verify the effectiveness of our DOCI model and the efficiency of the online OCI adjustment algorithm. Experimental results with a real‐world dataset show that DOCI achieves an improvement on system efficiency by up to 32%, compared with existing FT approaches.

A medical records and data capture and management system for Lassa fever in Sierra Leone: Approach, implementation, and challenges.

Situated in southeastern Sierra Leone, Kenema Government Hospital (KGH) maintains one of the world’s only Lassa fever isolation wards and was a strategic Ebola virus disease (EVD) treatment facility during the 2014 EVD outbreak. Since 2006, the Viral Hemorrhagic Fever Consortium (VHFC) has carried out research activities at KGH, capturing clinical and laboratory data for suspected cases of Lassa fever. Here we describe the approach, progress, and challenges in designing and maintaining a data capture and management system (DCMS) at KGH to assist infectious disease researchers in building and sustaining DCMS in low-resource environments. Results on screening patterns and case-fatality rates are provided to illustrate the context and scope of the DCMS covered in this study. A medical records system and DCMS was designed and implemented between 2010 and 2016 linking historical and prospective Lassa fever data sources across KGH Lassa fever units and its peripheral health units. Data were captured using a case report form (CRF) system, enzyme-linked immunosorbent assay (ELISA) plate readers, polymerase chain reaction (PCR) machines, blood chemistry analyzers, and data auditing procedures. Between 2008 and 2016, blood samples for 4,229 suspected Lassa fever cases were screened at KGH, ranging from 219 samples in 2008 to a peak of 760 samples in 2011. Lassa fever case-fatality rates before and following the Ebola outbreak were 65.5% (148/226) and 89.5% (17/19), respectively, suggesting that fewer, but more seriously ill subjects with Lassa fever presented to KGH following the 2014 EVD outbreak (p = .040). DCMS challenges included weak specificity of the Lassa fever suspected case definition, limited capture of patient survival outcome data, internet costs, lapses in internet connectivity, low bandwidth, equipment and software maintenance, lack of computer teaching laboratories, and workload fluctuations due to variable screening activity. DCMS are the backbone of international research efforts and additional literature is needed on the topic for establishing benchmarks and driving goal-based approaches for its advancement in developing countries.

Cognitive workload management on globally interoperable network of clouds

A new computing paradigm using Distributed Intelligent Managed Elements (DIME) and DIME Network Architecture (DNA) is used to demonstrate globally interoperable public and private cloud network deploying cloud agnostic workloads. The workloads are cognitive and able to adjust autonomously their structure and maintain desired quality-of-service (QoS). DNA is designed to provide a control architecture for workload self-management of non-functional requirements to address rapid fluctuations either in workload demand or fluctuations in available resources. Using DNA, a transaction intensive three-tier workload is migrated from a physical server to a virtual machine hosted in a public cloud without interrupting the service transactions. After migration, cloud agnostic inter-cloud and intra-cloud auto-scaling, auto- failover and live migration are demonstrated again, without disrupting the user experience or losing transactions.

Utilizing a Remote LED Stimulus to Concurrently Measure Cognitive and Visual Task Demand

The ISO 17488 standard Detection Response Task (DRT) has been validated as an effective tool for measuring fluctuations in cognitive workload while driving and performing secondary tasks. This research evaluated the possibility of consolidating a dual stimulus DRT to a single remote LED stimulus to concurrently measure visual and cognitive demand. Hit rate and reaction times to a remote LED stimulus and an ISO standard vibrotactor stimulus were compared for three in-vehicles tasks: a single task baseline, a cognitively demanding task, and a visually demanding task. Analyses showed that the remote LED and vibrotactor were equally sensitive to cognitive load, while the remote LED was more sensitive to visual load. We suggest the remote LED DRT system serves as a cost-effective, practical, sensitive method to concurrently assess cognitive and visual demand.

VMDFS: virtual machine dynamic frequency scaling framework in cloud computing

Development of modern techniques, such as virtualization, underlies new solutions to the problem of reducing energy consumption in cloud computing. However, for the infrastructure as a service providers, it would be a difficult process to guarantee energy saving. Analysis of the workload of applications shows that the average utilization of virtual machines has many fluctuations; therefore, deciding about how to control such fluctuations in virtual machines plays a significant role in improving the energy consumption of datacenters. In this study, an adaptable model called virtual machine dynamic frequency system (VMDFS) has been developed whose its innovation is monitoring the average fluctuations of workloads to vary the CPU frequency of virtual machines at runtime, dynamically. In this model, enhanced exponential moving average method is used to predict workload fluctuations, and then after calculating a smoothing coefficient for the utilization fluctuations, the coefficient is used to control the CPU frequency (or computing power) of virtual machines. The proposed model was compared with several base line approaches such as DVFS using real datasets from CoMon project (PlanetLab). The results of experiments on VMDFS show that besides the reduced service-level agreement violation by up to 43.22%, the overall energy consumption is reduced by 40.16%. In addition, the overall runtime before a host shutdown increased by 17.44% in average, while the runtime before a virtual machine migration increased by 7.2%. This also shows an overall decrease in the number of migrations.

Multi-prediction based scheduling for hybrid workloads in the cloud data center

Cloud computing can leverage over-provisioned resources that are wasted in traditional data centers hosting production applications by consolidating tasks with lower QoS and SLA requirements. However, the dramatic fluctuation of workloads with lower QoS and SLA requirements may impact the performance of production applications. Frequent task eviction, killing and rescheduling operations also waste CPU cycles and create overhead. This paper aims to schedule hybrid workloads in the cloud data center to reduce task failures and increase resource utilization. The multi-prediction model, including the ARMA model and the feedback based online AR model, is used to predict the current and the future resource availability. Decision to accept or reject a new task is based on the available resources and task properties. Evaluations show that the scheduler can reduce the host overload and failed tasks by nearly 70%, and increase effective resource utilization by more than 65%. The task delay performance degradation is also acceptable.

Recruitment Strategies for Cluster Randomized Controlled Trials Targeting Workplace Sedentary Behavior- a Retrospective Review

Increased demand for sedentary behavior reduction in workplace environments has led to the planning of large-scale interventions implemented at the group level in the form of cluster randomized controlled trials (RCTs). To date, limited evidence is available regarding cluster RCT recruitment strategies. PURPOSE: The purpose of this paper is to provide a review of recruitment strategies employed in a large cluster RCT targeting a reduction in workplace sedentary behavior. METHODS: Recruitment yields ([N enrolled/N screened] x 100) were calculated. Mean (±SD) and median worksite sizes were calculated at each recruitment step. The percentage of participants who progressed to each recruitment step (of the total N screened per worksite) was calculated to determine the mean percentage of a worksite successfully randomized. Recruitment barriers and modifications were recorded by the research team. A survey was completed by a subset of non-participants (N = 57) and thematic analyses conducted to examine reasons for non-participation, positive impacts and negative experiences. RESULTS: Cluster recruitment yield was 43% (24 worksites enrolled/56 screened). Individual recruitment yield was 49% (641 employees enrolled/1317 screened). On average, 52 ± 16% of the worksite was successfully randomized. Eighteen modifications were developed to overcome participant-related, context-related and research-related barriers. CONCLUSIONS: Researchers should plan to screen at least 200% of the intended number of worksites and they should target worksites that are approximately double the size of the intended cluster size to avoid loss of statistical power or timeline extensions. Acknowledging temporal fluctuations in worksite-specific workloads, providing options throughout the recruitment process, and adopting a participant-centered approach may facilitate cluster RCT success.

Providing Geo-Elasticity in Geographically Distributed Clouds

Geographically distributed cloud platforms are well suited for serving a geographically diverse user base. However, traditional cloud provisioning mechanisms that make local scaling decisions are not adequate for delivering the best possible performance for modern web applications that observe both temporal and spatial workload fluctuations. We propose GeoScale, a system that provides geo-elasticity by combining model-driven proactive and agile reactive provisioning approaches. GeoScale can dynamically provision server capacity at any location based on workload dynamics. We conduct a detailed evaluation of GeoScale on Amazon’s geo-distributed cloud and show up to 40% improvement in the 95th percentile response time when compared to traditional elasticity techniques.

Gloss

An important class of applications computes on long-running or infinite streams of data, often with known fixed data rates. The latter is referred to as synchronous data flow ~(SDF) streams. These stream applications need to run on clusters or the cloud due to the high performance requirement. Further, they require live reconfiguration and reoptimization for various reasons such as hardware maintenance, elastic computation, or to respond to fluctuations in resources or application workload. However, reconfiguration and reoptimization without downtime while accurately preserving program state in a distributed environment is difficult. In this paper, we introduce Gloss, a suite of compiler and runtime techniques for live reconfiguration of distributed stream programs. Gloss, for the first time, avoids periods of zero throughput during the reconfiguration of both stateless and stateful SDF based stream programs. Furthermore, unlike other systems, Gloss globally reoptimizes and completely recompiles the program during reconfiguration. This permits it to reoptimize the application for entirely new configurations that it may not have encountered before. All these Gloss operations happen in-situ, requiring no extra hardware resources. We show how Gloss allows stream programs to reconfigure and reoptimize with no downtime and minimal overhead, and demonstrate the wider applicability of it via a variety of experiments.

How can mindfulness be promoted? Workload and recovery experiences as antecedents of daily fluctuations in mindfulness.

While previous work on mindfulness has focused predominantly on the benefits of mindfulness and of mindfulness interventions, the present article addresses the question of how natural experiences of mindfulness can be promoted in the context of work. Accordingly, this article sheds light on day‐to‐day fluctuations in workload and recovery experiences (psychological detachment and sleep quality) as antecedents of state mindfulness. Furthermore, this study extends extant research that has documented beneficial effects of mindfulness on subsequent recovery experiences by arguing that the relationship between mindfulness and recovery experiences is reciprocal rather than unidirectional. Using an experience‐sampling design across five workdays and involving three daily measurement occasions, we found that sleep quality and workload were related to subsequent levels of mindfulness. While not displaying a significant direct relationship with mindfulness, psychological detachment was indirectly related to mindfulness via sleep quality. Fatigue was identified as an important mechanism explaining these relationships. Furthermore, findings confirmed that the relationship between mindfulness and recovery experiences is reciprocal rather than unidirectional. Taken together, this study contributes to an enriched understanding of the role of mindfulness in organizations by shedding light on factors that precede the experience of mindfulness and by pointing to the existence of gain spirals associated with recovery experiences and mindfulness.Practitioner points Organizations seeking to promote mindfulness among their workforce should try to keep workload to a manageable degree.Organizations may also pay attention to care for employees' day‐to‐day recovery as it has been shown to facilitate mindfulness.