User Service Level Agreement Research Articles

Performability evaluation of NoSQL-based storage systems

NoSQL database management systems (DBMS) have been commonly adopted in cloud storage applications, as they usually provide better performance and availability than relational DBMSs. Eventual consistency is a remarkable feature of many NoSQL database systems, in which not all redundant nodes have the newest data, but, eventually, such data will be present in all nodes. Distinct consistency levels can be utilized, but they may affect user experience and service level agreements. The assessment of consistency concerning the probability of accessing the newest data is not common, and the jointly evaluation of performance and availability is usually neglected. This work proposes a method based on stochastic Petri nets (SPN) for evaluating the consistency levels of storage systems based on NoSQL DBMS adopting quorum technique. The models take into account distinct consistency levels and redundant nodes for estimating system availability, throughput and the probability of accessing the newest data. Experimental results demonstrate the practical feasibility of our approach.

Energy‐efficient task scheduling and resource management in a cloud environment using optimized hybrid technology

AbstractCloud computing is a significant platform emerging for IT enterprises, business applications, and mobile computing. It is a dynamic environment where efficiently and properly allocating resources in such an environment becomes a tedious task. Most existing approaches cannot guarantee energy‐efficient resource management due to the existence of time‐dependent tasks. Cloud resources consume a huge amount of energy, which may reduce the Makespan of the entire network. The process of appropriate task scheduling may satisfy the user's requirement. Reducing energy consumption by satisfying the user's QoS requirements is essential to ensure each user's service level agreement (SLA). Therefore, to achieve an energy‐efficient and unique task scheduling, a technique for order preference by similarity to ideal solution (TOPSIS) based unique ranking (uRank‐TOPSIS) and a Hybrid state‐action‐reward‐state‐action (SARSA), Reinforcement Learning with Black widow Algorithm (HSRLBA) are proposed in this paper. Here, the task scheduling process is carried out using the uRank‐TOPSIS process by producing a set of unique weights and ranking the alternatives. Also, HSRLBA is used to perform resource allocation, which uses the Black widow Algorithm (BWA) to rapidly converge parallel agents in the SARSA model of RL. The Cloudsim simulator is used for experiments. The efficiency of the proposed framework is measured in terms of Makespan, Task Completion Ratio (TCR), total energy consumption, response time and resource utilization. The proposed uRank‐TOPSIS and HSRLBA achieved an energy consumption of 325 KWh, response time of 15.42 s, Makespan of the 1150 s, TCR of 98% and resource utilization of 92%. The proposed framework maximizes TCR and resource utilization and minimizes the Makespan, energy consumption, and response time.

Network Traffic Prediction: Apply the Transformer to Time Series Forecasting

Along with the development of technology and social progress, the Internet is increasingly widely used in life. Mobile communication, fiber optic broadband, and other essential Internet networks have gradually become indispensable in everyday life. The task of further improving and optimizing the quality of Internet network links and improving the efficiency of Internet networks has been on the agenda. This paper proposed a deep learning-based network traffic prediction model, which can capture the characteristics of network traffic information changes by inputting past network traffic data to achieve the effect of future network traffic prediction. The model structure is flexible and variable, which improves the problems of other methods that cannot capture long time series prediction features and cannot parallelize the output. It also has apparent advantages in time complexity and model convergence speed without the evident disadvantage of time lag. Based on this network traffic prediction model, it can help Internet service providers optimize network resource allocation, improve network performance, and allow Internet data centers to provide abnormal network warnings and improve user service level agreements.

Laying a Foundation for Digital Collections at the Property Information Resource Center

ABSTRACT Over a decade ago, the Harvard University Property Information Resource Center (PIRC) began digitizing its entire collection of more than a hundred thousand architectural drawings documenting the construction of the oldest university in the United States. Challenges and successes materialized throughout the project relating to the PIRC's mission, service level, and collection dependencies. Continuing to meet users' demanding needs while learning and revising best practices was ambitious yet ultimately achievable. In addition to producing high-quality images for digital preservation, secondary positive outcomes of the project were the conservation of drawings, improvements to the reference process, and the ability to expand these services beyond the traditional user group. To achieve the project goals, staff created a flexible workflow that ameliorated the condition of physical drawings in the collection while allowing them to uphold an established user service level agreement.

User preference–based QoS-aware service function placement in IoT-Edge cloud

In the Internet of Things-Edge cloud, service provision presents a challenge to operators to satisfy user service-level agreements while meeting service-specific quality-of-service requirements. This is because of inherent limitations in the Internet of Things-Edge in terms of resource infrastructure as well as the complexity of user requirements in terms of resource management in a heterogeneous environment like edge. An efficient solution to this problem is service orchestration and placement of service functions to meet user-specific requirements. This work aims to satisfy user quality of service through optimizing the user response time and cost by factoring in the workload variation on the edge infrastructure. We formulate the service function placement at the edge problem. We employ user service request patterns in terms of user preference and service selection probability to model service placement. Our framework proposal relies on mixed-integer linear programming and heuristic solutions. The main objective is to realize a reduced user response time at minimal overall cost while satisfying the user service requirements. For this, several parameters, and factors such as capacity, latency, workload, and cost constraints, are considered. The proposed solutions are evaluated based on different metrics and the obtained results show the gap between the heuristic user preference placement algorithm and the optimal solution to be minimal.

An energy‐efficient task‐scheduling algorithm based on a multi‐criteria decision‐making method in cloud computing

SummaryThe massive growth of cloud computing has led to huge amounts of energy consumption and carbon emissions by a large number of servers. One of the major aspects of cloud computing is its scheduling of many task requests submitted by users. Minimizing energy consumption while ensuring the user's QoS preferences is very important to achieving profit maximization for the cloud service providers and ensuring the user's service level agreement (SLA). Therefore, in addition to implementing user's tasks, cloud data centers should meet the different criteria in applying the cloud resources by considering the multiple requirements of different users. Mapping of user requests to cloud resources for processing in a distributed environment is a well‐known NP‐hard problem. To resolve this problem, this paper proposes an energy‐efficient task‐scheduling algorithm based on best‐worst (BWM) and the Technique for Order Preference by Similarity to Ideal Solution (TOPSIS) methodology. The main objective of this paper is to determine which cloud scheduling solution is more important to select. First, a decision‐making group identify the evaluation criteria. After that, a BWM process is applied to assign the importance weights for each criterion, because the selected criteria have varied importance. Then, TOPSIS uses these weighted criteria as inputs to evaluate and measure the performance of each alternative. The performance of the proposed and existing algorithms is evaluated using several benchmarks in the CloudSim toolkit and statistical testing through ANOVA, where the evaluation metrics include the makespan, energy consumption, and resource utilization.

Control strategies for adaptive resource allocation in cloud computing

Using a compute infrastructure efficiently to execute jobs while respecting Service Level Agreements (SLAs) and thereby guaranteeing Quality of Service (QoS) poses a number of challenges. One such challenge lies in the fact that SLAs are set prior to the execution of a job, but the execution environment is subject to a number of possible disturbances, such as poor knowledge about actual resource necessity, demand peaks and hardware malfunctions, amongst others. Thus by using a fixed resource allocation, the manager of a shared computing environment risks violating user SLAs. Furthermore, the complexity of managing several workload executions increases with the number of workloads, implying the need for an automatic method to manage and control the execution of workloads. The execution time SLA is specially important in streaming scenarios such as web applications and continuous video processing, and is the focus of this paper. A method based on adaptive model predictive control (aMPC) is proposed here to adapt the amount of allocated resources to iterative workloads. The methodology is tested applied to Deep Learning Workloads, in standalone and multi-workload versions. The results show that using adaptive optimal control with a linearized model improves performance with respect to simpler control laws as well as reinforcement learning approaches.

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.

Evaluating the cooling and computing energy demand of a datacentre with optimal server provisioning

As cloud computing continues to gain significance across fields, the energy consumption of datacentres creates new challenges in the design and operation of computer systems, with cooling remaining a key part of the total energy expenditure. We investigate the implications of increasing the room temperature setpoint in datacentres to save energy. For this, we develop a holistic model for the energy consumption of the server room that depends on user workload and service level agreement constraints, and that considers both cooling and computing energy dissipation. The model is applicable to a steady-state analysis of the system and brings insight into the impact of the most relevant parameters that affect the net energy consumption, such as the outside temperature, room temperature setpoint, and user demand. We analyse both static and dynamic server provisioning cases. In the latter case, a global power management scheme determines the optimal number of servers required to handle the incoming user demand to fulfil a target service level objective. Finally, we consider the extra energy needed to maintain service continuity under the expected higher server mortality rate due to warmer operational temperatures. Energy and temperature measurements acquired from a server machine running scientific benchmark programs allow to realistically fix model parameters for the study and to obtain pragmatic conclusions.

Operational cost minimization of distributed data centers through the provision of fair request rate allocations while meeting different user SLAs

Data centers as computing infrastructures for cloud services have been growing in both number and scale. However, they usually consume enormous amounts of electricity that incur high operational costs of cloud service providers. Minimizing these operational costs thus becomes one main challenge in cloud computing. In this paper, we study the operational cost minimization problem in a distributed cloud computing environment that not only considers fair request rate allocations among web portals but also meets various Service Level Agreements (SLAs) between users and the cloud service provider, with an objective to maximize the number of user requests admitted while keeping the operational cost minimized, by exploiting the electricity diversity. To this end, we first propose an adaptive operational cost optimization framework that incorporates time-varying electricity prices and dynamic user request rates. We then devise a fast approximation algorithm with a provable approximation ratio for the problem, by utilizing network flow techniques. Finally, we evaluate the performance of the proposed algorithm through experimental simulations, using real-life electricity price data sets. Experimental results demonstrate that the proposed algorithm is very promising, and the solution obtained is nearly optimal.

Adaptive Registration in TWDM-PON With ONU Migrations

In time and wavelength division multiplexed passive optical networks (TWDM-PONs), optical network unit (ONU) migration (equivalently retuning working wavelength pairs) is an effective approach to reducing power consumption at the optical line terminal (OLT) side. Migration delay, the time from when a migration is initiated until the migrating ONU(s) have successfully joined the target wavelength pair, may affect the user service level agreement (SLA) and should be minimized. After an ONU migration process has been triggered, both the ONUs to be migrated (the migrating ONUs) and the newly incoming ones will have to register (referred to as activation in the ITU-T G.984.3 standard) in the upcoming quiet windows. This makes the number of ONUs competing in a single quiet window much more dynamic and leads to unpredictable registration performance and longer registration delay, which in turn will have a negative impact on the migration delay. In this paper, we propose an adaptive registration scheme in a TWDM-PON with ONU migrations. In the proposed scheme, the size of the quiet window is adaptively adjusted according to the up-to-date number of involved ONUs (including both ONUs to be migrated and the newly incoming ones). We show through simulations that the proposed scheme reduces average migration delay and increases the bandwidth utilization, especially when a large number of ONUs are involved in a registration process. In addition, we further investigate the length of the discovery cycle and find a shorter discovery cycle is needed to reduce migration delay and satisfy the requirement of the ONU SLA for a large number of involved ONUs. To realize a 99.999% ONU SLA, the length of the discovery cycle has to be shorter than 0.1 s for 128 involved ONUs and 0.15 s for 32 involved ONUs.

ONU migration in dynamic Time and Wavelength Division Multiplexed Passive Optical Network (TWDM-PON)

We propose a novel Optical Network Units (ONU) migration mechanism within the Time and Wavelength Division Multiplexed PON (TWDM-PON) framework by rearranging the working ONUs to a minimum number of wavelengths and turning off the rest to save power. We show through simulation that the number of working wavelengths can be minimized up to a theoretical lower bound, e.g. 23%, under the typical ONU online profiles. We further investigate how the migration interval and delay influence the user Service Level Agreement (SLA). We find that under the example ONU online profiles, 99.99% ONU SLA can be realized with the migration delay of 1000 milli-seconds. However to realize 99.999% ONU SLA, the migration delay must be 100 milli-seconds or lower.

IMeter: An integrated VM power model based on performance profiling

The unprecedented burst in power consumption encountered by contemporary datacenters continually boosts the development of energy efficient techniques from both hardware and software perspectives to alleviate the energy problem. The most widely adopted power saving solutions in datacenters that deliver cloud computing services are power capping and VM consolidation. However, without the capability to track the VM power usage precisely, the combined effect of the above two techniques could cause severe performance degradation to the consolidated VMs, thus violating the user service level agreements. In this paper, we propose an integrated VM power model called iMeter, which overcomes the drawbacks of overpresumption and overapproximation in segregated power models used in previous studies. We leverage the kernel-based performance counters that provide accurate performance statistics as well as high portability across heterogeneous platforms to build the VM power model. Principal component analysis is applied to identify performance counters that show strong impact on the VM power consumption with mathematical confidence. We also present a brief interpretation of the first four selected principal components on their indications of VM power consumption. We demonstrate that our approach is independent of underlying hardware and virtualization configurations with clustering analysis. We utilize the support vector regression to build the VM power model predicting the power consumption of both a single VM and multiple consolidated VMs running various workloads. The experimental results show that our model is able to predict the instantaneous VM power usage with an average error of 5% and 4.7% respectively against the actual power measurement.

GRA Based Network Selection in Heterogeneous Wireless Networks

4G wireless networks will integrate heterogeneous technologies such as Wireless LAN and third generation (3G) cellular networks and have the capability to offer various services at any time as per user requirements, anywhere with seamless interoperability at affordable cost. One important challenge in such a heterogeneous wireless environment is to enable network selection mechanisms in order to keep the mobile users always best connected anywhere and at any time. In this paper, a multi-criteria access network selection algorithm is proposed in Worldwide Interoperability for Microwave Access---Wireless Fidelity environment, in order to facilitate the provision of high quality services and at the same time to satisfy different types of user service level agreements. Analytical hierarchy process (AHP) and grey relational analysis (GRA) methods are applying for optimal access network selection. The proposed methodology combines the AHP to decide the relative weights of criteria set according to network's performance, as well as the GRA to rank the network alternatives. The advantages of the GRA method are that the results are based on the original data, the calculations are simple and straightforward, and finally it is one of the best methods to make decision under heterogeneous wireless network environment.

A Stackelberg game to derive the limits of energy savings for the allocation of data center resources

Energy-related costs are becoming one of the largest contributors to the overall cost of operating a data center, whereas the degree of data center utilization continues to be very low. An energy-aware dynamic provision of resources based on the consolidation of existing application instances can simultaneously address the underutilization of servers while greatly reducing energy costs. The economics behind energy costs cannot be treated separately from resource provision and allocation. However, current scheduling techniques based on market mechanisms do not specifically deal with such a scenario. To establish the upper bound of energy savings, we model the problem of minimizing energy consumption when allocating resources to networked applications as a Stackelberg leadership game. The model is applied to a proportional-share mechanism in which resource providers can maximize profit by minimizing energy costs, while users can select resources that ensure that their minimum requirements are satisfied. We show that our mechanism can determine the optimal set of resources switched on and off, while maintaining user service level agreements (SLAs) — even in realistic conditions considering incomplete information.

Modeling and Performance Analysis to Predict the Behavior of a Divisible Load Application in a Cloud Computing Environment

Cloud computing is an emerging technology where IT resources are virtualized to users as a set of a unified computing resources on a pay per use basis. The resources are dynamically chosen to satisfy a user Service Level Agreement and a required level of performance. Divisible load applications occur in many scientific and engineering applications and can easily be mapped to a Cloud using a master-worker pattern. However, those applications pose challenges to obtain the required performance. We model divisible load applications tasks processing on a set of cloud resources. We derive a novel model and formulas for computing the blocking probability in the system. The formulas are useful to analyze and predict the behavior of a divisible load application on a chosen set of resources to satisfy a Service Level Agreement before the implementation phase, thus saving time and platform energy. They are also useful as a dynamic feedback to a cloud scheduler for optimal scheduling. We evaluate the model in a set of illustrative scenarios.

Implementation and performance evaluation of a distributed conjugate gradient method in a cloud computing environment

SUMMARYCloud computing is an emerging technology where information technology resources are provisioned to users in a set of a unified computing resources on a pay per use basis. The resources are dynamically chosen to satisfy a user service level agreement and a required level of performance. A cloud is seen as a computing platform for heavy load applications. Conjugate gradient (CG) method is an iterative linear solver that is used by many scientific and engineering applications to solve a linear system of algebraic equations. CG generates a heavy load of computation, and therefore, it slows the performance of the applications using it. Distributing CG is considered as a way to increase its performance. However, running a distributed CG, based on a standard API, such as Message Passing Interface, in a cloud face many challenges, such as the cloud processing and networking capabilities. In this work, we present an in‐depth analysis of the CG algorithm and its complexity to develop adequate distributed algorithms. The implementation of these algorithms and their evaluation in our cloud environment reveal the gains and losses achieved by distributing the CG. The performance results show that despite the complexity of the CG processing and communication, a speedup gain of at least 1157.7 is obtained using 128 cores compared with National Aeronautics and Space Administration Advanced Supercomputing sequential execution. Given the emergence of clouds, the results in this paper analyzes performance issues when a generic public cloud, along with a standard development library, such as Message Passing Interface, is used for high‐performance applications, without the need of some specialized hardware and software. Copyright © 2012 John Wiley & Sons, Ltd.

A Systematic Review and Assessment of Aspect-oriented Methods Applied to Business Process Adaptation

Today's ever-changing business environments, comprised among other things of customer expectations, market demands, and legal obligations, require dynamic and adaptive business processes. Hence, enterprises need to monitor and improve their business processes against their business goals and constraints. Aspect-oriented development is known to have helped designers cope with changing con- cerns in software, even dynamically. In this paper, we per- form a systematic literature review of aspect-oriented approaches for business process adaptation. We observe that current methods focus on i) composing and swapping services based on Quality of Service (QoS), cost, rules, poli- cies, and constraints, as well as in the event of failure, ii) extracting roles and crosscutting concerns from composite services, iii) customizing process instances based on user profiles or Service Level Agreements, iv) adapting service composition and collaboration policies, and v) using moni- toring aspects to detect undesired situations. This review also suggests that our own aspect-oriented process modeling and adaptation framework is novel because none of the other approaches considers organization goals, performance and constraints as a whole when improving business proc- esses. In addition, given much prior research on aspect- oriented service composition is available, we are confident that our modeling framework is realizable.

Taking Turns With Adaptive Cycle Time a Decentralized Media Access Scheme for LR-PON

The extended network span of next-generation long-reach passive optical networks (LR-PONs) results in extremely long propagation delays that severely degrade the performance of centralized bandwidth allocation algorithms. This is because these algorithms are based on bandwidth negotiation messages frequently exchanged between the optical line terminal (OLT) in the central office and optical network units (ONUs) near the users, which become seriously delayed when the network is extended. To address this problem, we propose a decentralized media access scheme for emerging LR-PONs to make the performance independent of the physical length between the OLT and ONUs. We also maintain centralized control over the network, usually missed in conventional decentralized schemes, to support and manage quality of service according to user service level agreements. The scheme thus combines decentralized media access with centralized control to meet the special requirements of emerging LR-PONs. We investigate the performance of the proposed scheme in contrast with centralized schemes under worst case conditions. We also explore various approaches to further enhance the performance of our scheme. Simulation results show that the average upstream packet delay can be decreased by 60% while also maintaining a high throughput.

Network selection in a WiMAX–WiFi environment

The tremendous growth of wireless technologies has introduced the potential of continuous service adaptation to the users’ needs by giving them the ability to be able to select and access the proper network based on different criteria. Moreover, next generation wireless networks have been designed to provide support for multimedia services, with different traffic characteristics and different Quality of Service (QoS) guarantees. However, the expansion of Wireless Local Area Networks (WLANs) and Worldwide Interoperability for Microwave Access (WiMAX) networks poses new research era in the decision of the access network selection. In this paper, the existing access network selection schemes are classified into three categories: the network-centric, the user-centric and the collaborative schemes, and are analyzed respectively. Moreover, we propose a multicriteria access network selection algorithm applied in a WiMAX–WLAN environment, in order to facilitate the provision of high quality services and at the same time to satisfy different types of user Service Level Agreements (SLAs).