Auto-scaling Method Research Articles

Proactive auto-scaling technique for web applications in container-based edge computing using federated learning model

Edge computing has emerged as an attractive alternative to traditional cloud computing by utilizing processing, network, and storage resources close to end devices, such as Internet of Things (IoT) sensors. Edge computing is still in its infancy, and resource provisioning and service scheduling remain research concerns. Kubernetes is a container orchestration tool in distributed environments. Proactive auto-scaling techniques in Kubernetes improve utilization by allocating resources based on future workload prediction. However, prediction models run on central cloud nodes, necessitating data transfer between edge and cloud nodes, which increases latency and response time. We present FedAvg-BiGRU, a proactive auto-scaling method in edge computing based on FedAvg and multi-step prediction by a Bidirectional Gated Recurrent Unit (BiGRU). FedAvg is a technique for training machine learning models in a Federated Learning (FL) model. FL reduces network traffic by exchanging only model updates rather than raw data, relieving learning models of the need to store data on a centralized cloud server. In addition, a technique for determining the number of Kubernetes pods based on the Cool Down Time (CDT) concept has been developed, preventing contradictory scaling actions. To our knowledge, our work is the first to employ FL for proactive auto-scaling in cloud and edge computing. The results demonstrate that the FedAvg-BiGRU method has a slightly higher prediction error than the load centralized processing mode, although this difference is not statistically significant. At the same time, it reduces the amount of data transmission between the edge nodes and the cloud server.

A Q-learning based auto-scaling approach for provisioning big data analysis services in cloud environments

Currently, analyzing big data to capture hidden values in various fields is one of the most popular research directions. Analytics-as-a-Service (AaaS) providers typically construct a common platform by renting virtual machine (VM) resources from Infrastructure-as-a-Service (IaaS) providers instead of owning their own physical resources, to provision big data analysis services to end users. However, due to the fact that service demands from AaaS users tend to fluctuate over time in reality, how to dynamically adjust the type and quantity of VMs rented from IaaS providers has become an urgent problem to be solved. In this article, we assume that IaaS providers can offer VM resources in two charging modes, and service demand workloads arrive stochastically. We formulate the problem as a Markov decision process and propose a Q-Learning based auto-scaling method, which explores the trade-off between AaaS providers’ cost and VM resource utilization. Eventually, we evaluate our method on both real traces and simulated traces and compare it with some existing methods. Experimental results demonstrate that our method can not only achieve the goal of reducing cost, but also ensure the high utilization of VM resources in the long run.

Intent-driven orchestration of serverless applications in the computing continuum

Following the emergence of Internet of Things (IoT) and edge computing technologies and the development of distributed applications with strict Quality of Service (QoS) requirements, a transition is underway from centralized computing models to distributed computing continuum systems. The latter provide abstractions of the available resources, while they lead to the development of orchestration mechanisms with increased distributed intelligence and autonomy characteristics. These characteristics make the computing continuum suitable for managing serverless computing applications, considering their loose coupling with the resources and the need for reactiveness and efficiency to serve dynamic workloads in the edge and cloud part of the infrastructure. This paper presents an intent-based approach for the orchestration of a function chain over an edge and cloud infrastructure. We propose a Reinforcement Learning (RL) based autoscaling method that incorporates a reward function that aims to provide optimal utilization of resources and timely autoscaling of the functions, taking advantage of methods for workload prediction. Based on a high-level intent, the proposed optimization problem dictates the scheduling of the functions, minimizing the communication overhead and transformation cost while also considering the energy efficiency of the infrastructure. The proposed solution is evaluated with other state-of-the-art techniques for autoscaling and scheduling of serverless functions in a small edge infrastructure. Our solution provides 1.52% QoS violations with a slight increase in the deployed resources, while also significantly reducing the total cost and power consumption for the deployment of the function chain.

A Trend Detection-Based Auto-Scaling Method for Containers in High-Concurrency Scenarios

A Trend Detection-Based Auto-Scaling Method for Containers in High-Concurrency Scenarios

Toward Optimal Load Prediction and Customizable Autoscaling Scheme for Kubernetes

Most enterprise customers now choose to divide a large monolithic service into large numbers of loosely-coupled, specialized microservices, which can be developed and deployed separately. Docker, as a light-weight virtualization technology, has been widely adopted to support diverse microservices. At the moment, Kubernetes is a portable, extensible, and open-source orchestration platform for managing these containerized microservice applications. To adapt to frequently changing user requests, it offers an automated scaling method, Horizontal Pod Autoscaler (HPA), that can scale itself based on the system’s current workload. The native reactive auto-scaling method, however, is unable to foresee the system workload scenario in the future to complete proactive scaling, leading to QoS (quality of service) violations, long tail latency, and insufficient server resource usage. In this paper, we suggest a new proactive scaling scheme based on deep learning approaches to make up for HPA’s inadequacies as the default autoscaler in Kubernetes. After meticulous experimental evaluation and comparative analysis, we use the Gated Recurrent Unit (GRU) model with higher prediction accuracy and efficiency as the prediction model, supplemented by a stability window mechanism to improve the accuracy and stability of the prediction model. Finally, with the third-party custom autoscaling framework, Custom Pod Autoscaler (CPA), we packaged our custom autoscaling algorithm into a framework and deployed the framework into the real Kubernetes cluster. Comprehensive experiment results prove the feasibility of our autoscaling scheme, which significantly outperforms the existing Horizontal Pod Autoscaler (HPA) approach.

Proactive Random-Forest Autoscaler for Microservice Resource Allocation

Cloud service providers have been shifting their workloads to microservices to take advantage of their modularity, flexibility, agility, and scalability. However, numerous obstacles remain to achieving the most out of microservice deployments, especially in terms of a Quality of Service (QoS). One possible approach to overcoming these obstacles is to perform autoscaling, which is the ability of cloud infrastructure and services to scale themselves up or down by changing their resource pool. There are two major categories of autoscaling: reactive and proactive. In reactive autoscaling, a feedback loop based on current workload resource usage is implemented to guide resource scaling. One disadvantage of reactive autoscaling is that it may result in inconsistencies between workload demand and resource allocation. In proactive autoscaling, a prediction model is used to guide the future allocation of resources according to current workload metrics. In this paper, a novel proactive autoscaling method is introduced where a two-state, machine-learning Random Forest (RF) model is designed to forecast the future CPU and memory utilization values required by the microservice workload. These predicted values are then used to adjust the resource pool both vertically (hardware resources) and horizontally (microservice replicas). The RF proactive autoscaler has been implemented on a home-grown, open-source microservice prototyping platform and verified using real-world workloads. The experiments show that the RF proactive autoscaler outperforms state-of-the-art ones in terms of allocated resources and latency. The increase in the utilization of allocated resources can reach 90% and the improvement in end-to-end latency, measured by the <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$95^{th}$ </tex-math></inline-formula> percentile, can reach 95%.

KSN: Modeling and simulation of knowledge using machine learning in NFV/SDN-based networks

The rapid advancement of networking and computing technologies has led to the emergence of a wide range of diverse dynamic network services that present networks with new challenges, including in the provision and management of services. Recently, SDN and NFV technologies can be effective solutions in the dynamic management of network functions. For dealing with dynamic and complex problems in the field of NFV and SDN issues, the use of knowledge-based systems can be a good solution that has received more and more attention from researchers and developers. Despite the increasing attention to the use of knowledge such as machine learning and deep learning methods in the NFV and SDN field, the lack of a simulation framework with the ability to knowledge concept to implement and evaluate new methods has created a gap. In order to evaluate a new method on real platforms, a large-scale infrastructure has to be used, which faces challenges such as availability, cost, time and implementation difficulties. Since the existing simulation frameworks for networks using NFV and SDN technologies (e.g. CloudsimSDNNFV) do not support the knowledge-based concept, to fill this gap, we propose a highly extensible, interoperable, and scalable simulation framework for developing knowledge-based methods such as machine learning algorithms in the field of NFV and SDN, such as virtualized network functions auto-scaling, placement and migration. Also, for the proposed Knowledge-defined SDN/NFV (KSN) framework, an architecture is provided by defining the knowledge plane in the SDN/NFV integration structure. In addition, a new auto-scaling method based on the prediction of NFV resources consumption using machine learning has been proposed and simulated to demonstrate the capabilities of the framework and its evaluation. As we know, this is the first machine learning-based method in the NFV field, including VNF auto-scaling, which has been simulated in a network environment, and the impact of using machine learning has been practically evaluated and observed. The evaluation results show that the proposed method has a better impact on the SLA violation rate and resource usage than ElasticSFC and the basic version (without any auto-scaling method) and therefore it can be concluded that resource prediction is an effective and efficient solution for earlier decision making and VNF auto-scaling.

Burst-Aware Predictive Autoscaling for Containerized Microservices

Autoscaling methods are used for cloud-hosted applications to dynamically scale the allocated resources for guaranteeing Quality-of-Service (QoS). The public-facing application serves dynamic workloads, which contain bursts and pose challenges for autoscaling methods to ensure application performance. Existing State-of-the-art autoscaling methods are burst-oblivious to determine and provision the appropriate resources. For dynamic workloads, it is hard to detect and handle bursts online for maintaining application performance. In this article, we propose a novel burst-aware autoscaling method which detects burst in dynamic workloads using workload forecasting, resource prediction, and scaling decision making while minimizing response time service-level objectives (SLO) violations. We evaluated our approach through a trace-driven simulation, using multiple synthetic and realistic bursty workloads for containerized microservices, improving performance when comparing against existing state-of-the-art autoscaling methods. Such experiments show an increase of <inline-formula><tex-math notation="LaTeX">$\times $</tex-math><alternatives><mml:math><mml:mo>&#x00D7;</mml:mo></mml:math><inline-graphic xlink:href="iqbal-ieq1-2995937.gif"/></alternatives></inline-formula>1.09 in total processed requests, a reduction of <inline-formula><tex-math notation="LaTeX">$\times $</tex-math><alternatives><mml:math><mml:mo>&#x00D7;</mml:mo></mml:math><inline-graphic xlink:href="iqbal-ieq2-2995937.gif"/></alternatives></inline-formula>5.17 for SLO violations, and an increase of <inline-formula><tex-math notation="LaTeX">$\times $</tex-math><alternatives><mml:math><mml:mo>&#x00D7;</mml:mo></mml:math><inline-graphic xlink:href="iqbal-ieq3-2995937.gif"/></alternatives></inline-formula>0.767 cost as compared to the baseline method.

Predictive Hybrid Autoscaling for Containerized Applications

One of the main challenges in deploying container service is providing the scalability to satisfy the service performance and avoid resource wastage. To deal with this challenge, Kubernetes provides two kinds of scaling mode: vertical and horizontal. Several existing autoscaling methods make efforts to improve the default autoscalers in Kubernetes; however, most of these works only focus on one scaling mode at the same time, which results in some limitations. Only horizontal scaling may lead to low utilization of containers due to the fixed amount of resources for each instance, especially in the low-request period. In contrast, only vertical scaling may not ensure the quality of service (QoS) requirements in case of bursty workload due to reaching the upper limit. Besides, it is also necessary to provide burst identification for auto-scalers to guarantee service performance. This paper proposes a hybrid autoscaling method with burst awareness for containerized applications. This new approach considers a combination of both vertical and horizontal abilities to satisfy the QoS requirement while optimizing the utilization of containers. Our proposal uses a predictive method based on the machine learning technique to predict the future demand of the application and combines it with a burst identification module, which makes scaling decisions more effective. Experimental results show an enhancement in maintaining the response time below the QoS constraint whereas remaining high utilization of the deployment compared with existing baseline methods in single scaling mode.

Predictive Auto-Scaling of Multi-Tier Applications Using Performance Varying Cloud Resources

The performance of the same type of cloud resources, such as virtual machines (VMs), varies over time mainly due to hardware heterogeneity, resource contention among co-located VMs, and virtualization overhead. The performance variation can be significant, introducing challenges to learn workload-specific resource provisioning policies to automatically scale the cloud-hosted applications to maintain the desired response time. Moreover, auto-scaling multi-tier applications using minimal resources is even more challenging because bottlenecks may occur on multiple tiers concurrently. In this paper, we address the problem of using performance varying VMs for gracefully auto-scaling a multi-tier application using minimal resources to handle dynamically increasing workloads and satisfy the response time requirements. The proposed system uses a supervised learning method to identify the appropriate resources provisioning for multi-tier applications based on the prediction of the application response time and the request arrival rate. The supervised learning method learns a state transition configuration map which encodes a resource allocation states invariant to the underlying VMs performance variations. This configuration map helps to use performance varying resources in predictive autoscaling method. Our experimental evaluation using a real-world multi-tier web application hosted on a public cloud shows an improved application performance with minimal resources compared to conventional predictive auto-scaling methods.

Predictive Autoscaling of Microservices Hosted in Fog Microdata Center

Fog computing provides microdata center (MDC) facilities closer to the users and applications, which help to overcome the application latency and response time concerns. However, guaranteeing specific service-level objectives (SLOs) for the applications running on the MDC requires automatic scaling of allocated resources by efficiently utilizing the available infrastructure capacity. In this article, we propose a novel predictive autoscaling method for microservices running on the fog MDC to satisfy the application response time SLO. Initially, our proposed approach uses a reactive rule-based autoscaling method to gather the training dataset for building the predictive autoscaling model. The proposed approach is efficient, as it can learn the predictive autoscaling model using an increasing synthetic workload. The learned predictive autoscaling model is used to manage the application resources serving different realistic workloads effectively. Our experimental evaluation using two synthetic and three realistic workloads for two benchmark microservice applications on a real MDC shows excellent performance compared to the existing state-of-the-art baseline rule-based autoscaling method. The proposed autoscaling method yields 75.51% reduction in the number of rejected requests and 77.53% fewer number of SLO violations compared to the baseline autoscaling methods by using only 9.20% additional data center resources at the fog layer.

Auto-Scaling Network Service Chains Using Machine Learning and Negotiation Game

Network Function Virtualization (NFV) enables Network Operators (NOs) to efficiently respond to the increasing dynamicity of network services. Virtual Network Functions (VNFs) running on commercial off-the-shelf servers are easy to deploy, update, monitor, and manage. Such virtualized services are often deployed as Service Chains (SCs), which require in-sequence placement of computing and memory resources as well as routing of traffic flows. Due to the ongoing migration towards cloudification of networks, the concept of auto-scaling which originated in Cloud Computing, is now receiving attention from networks professionals too. Prior studies on auto-scaling use measured load to dynamically react to traffic changes. Moreover, they often focus on only one of the resources (e.g., compute only, or network capacity only). In this study, we consider three different resource types: compute, memory, and network bandwidth. In prior studies, NO takes auto-scaling decisions, assuming tenants are always willing to auto-scale, and Quality of Service (QoS) requirements are homogeneous. Our study proposes a negotiation-game-based auto-scaling method where tenants and NO both engage in the auto-scaling decision, based on their willingness to participate, heterogeneous QoS requirements, and financial gain (e.g., cost savings). In addition, we propose a proactive Machine Learning (ML) based prediction method to perform SC auto-scaling in dynamic traffic scenario. Numerical examples show that our proposed SC auto-scaling methods powered by ML present a win-win situation for both NO and tenants (in terms of cost savings).

Key influencing factors of the Kubernetes auto-scaler for computing-intensive microservice-native cloud-based applications

Nowadays, different types of computing-intensive services such as mechanical, aerospace, civil and environmental applications are often deployed on the cloud since it offers a convenient on-demand model for renting resources and easy-to-use elastic infrastructures. Moreover, the modern software engineering disciplines exploit orchestration tools such as Kubernetes to run cloud applications based on a set of microservices packaged in containers. On the one hand, in order to ensure the users’ experience, it is necessary to allocate enough number of container instances before the workload intensity surges at runtime. On the other hand, renting expensive cloud-based resources can be unaffordable over a long period of time. Therefore, the choice of a reactive auto-scaling method may significantly affect both response time and resource utilisation. This paper presents a set of key factors which should be considered in the development of auto-scaling methods. Through a set of experiments, a discussion follows to help shed light on how such factors influence the performance of auto-scaling methods under different workload conditions such as on-and-off, predictable and unpredictable bursting workload patterns. Due to suitable results, the proposed set of key factors are exploited in the PrEstoCloud software system for microservice-native cloud-based computationally-intensive applications.

Dynamic Autoselection and Autotuning of Machine Learning Models for Cloud Network Analytics

Cloud network monitoring data is dynamic and distributed. Signals to monitor the cloud can appear, disappear or change their importance and clarity over time. Machine learning (ML) models tuned to a given data set can therefore quickly become inadequate. A model might be highly accurate at one point in time but may lose its accuracy at a later time due to changes in input data and their features. Distributed learning with dynamic model selection is therefore often required. Under such selection, poorly performing models (although aggressively tuned for the prior data) are retired or put on standby while new or standby models are brought in. The well-known method of Ensemble ML (EML) may potentially be applied to improve the overall accuracy of a family of ML models. Unfortunately, EML has several disadvantages, including the need for continuous training, excessive computational resources, requirement for large training datasets, high risks of overfitting, and a time-consuming model-building process. In this paper, we propose a novel cloud methodology for automatic ML model selection and tuning that automates model building and selection and is competitive with existing methods. We use unsupervised learning to better explore the data space before the generation of targeted supervised learning models in an automated fashion. In particular, we create a Cloud DevOps architecture for autotuning and selection based on container orchestration and messaging between containers, and take advantage of a new autoscaling method to dynamically create and evaluate instantiations of ML algorithms. The proposed methodology and tool are demonstrated on cloud network security datasets.

Abstract P6-13-05: Association between the UGT2B17 gene deletion, exemestane metabolites and vasomotor QOL in women participating on the MAP3 prevention trial

Abstract Background: The aromatase inhibitor Exemestane (EXE) reduces the risk of breast cancer in postmenopausal women. However, participants have varied responses to EXE treatment in terms of efficacy and toxicity, possibly due to differences in EXE metabolism. One of the main elimination pathways for EXE is through glucuronidation by UGT2B17. Aims: This project examined the relationship between the UGT2B17 gene deletion, EXE metabolites and menopause-related quality of life (QOL) in postmenopausal women. Hypothesis: Glucuronidation of the main EXE metabolite, 17-dihydroexemestane (17-DHE), is reduced in women with the UGT2B17 double gene deletion, leading to increased circulating 17-DHE and potential toxicity. Methods: This study included 3576 women nested within the CCTG MAP.3 trial, who were allocated to EXE or placebo treatment groups. Genotyping analysis was conducted with baseline blood cell DNA using real-time PCR and allelic discrimination. Women who were homozygous null were considered “exposed”. In addition to EXE, EXE metabolites including 17-DHE and glucuronidated 17-DHE (17 DHE-Gluc) were analyzed from serum by UPLC/MS. Ratios of the main metabolites (17-DHE/EXE) and glucuronidated metabolites (17-DHE-Gluc/17-DHE) were standardized, using an autoscaling method. Metabolite levels that were below the detection limit were replaced by “half the detection limit for that metabolite”. Women had the outcome if they experienced a clinically meaningful (&amp;gt;10%) worsening in vasomotor QOL from baseline within the first year. Modified Poisson regression models were used to calculate the relative risks for both the (1) UGT2B17 gene deletion and (2) metabolite ratios and vasomotor QOL. Results: Ten percent of participants exhibited the homozygous UGT2B17 deletion genotype. There was no significant relationship between the UGT2B17 deletion polymorphism and worsened vasomotor QOL (RR= 1.04, 95% CI: 0.93, 1.17), adjusted for age, race and treatment. Among women with no vasomotor symptoms at baseline but extremely bothersome symptoms at follow-up (incident vasomotor symptoms), there was a suggestive but non-significant protective effect of the UGT2B17 deletion (RR=0.61, 95% CI: 0.32-1.19). This effect was more extreme in the placebo arm (RR=0.20) than in the EXE arm (RR=0.78; p-interaction=0.17). Among women on EXE, levels of EXE and 17-DHE were not different between UGT2B17 genotypes, but levels of 17-DHE-Gluc were significantly lower for the UGT2B17 deletion genotype (p=&amp;lt;0.0001). An increasing ratio of 17-DHE-Gluc/17-DHE [per standard deviation (SD) increase] had a borderline protective effect against worsened vasomotor QOL (RR=0.94, p=0.049), adjusted for age and race. In contrast, an increasing ratio of 17-DHE /EXE (per SD increase) was associated with a small but significant increased risk of worsened vasomotor QOL (RR=1.02, p=0.01). The effect observed for the 17-DHE/EXE ratio was stronger for very bothersome incident vasomotor symptoms at follow-up, but this did not reach statistical significance [17-DHE/EXE (per SD increase): RR=1.36, p=0.12]. Conclusion: EXE metabolite levels could potentially be used as a biomarker for extreme vasomotor QOL changes in breast cancer chemoprevention settings using EXE. Citation Format: Richardson H, Knight B, Chen G, Luo S, Massey T, Goss PE, Lazarus P. Association between the UGT2B17 gene deletion, exemestane metabolites and vasomotor QOL in women participating on the MAP3 prevention trial [abstract]. In: Proceedings of the 2018 San Antonio Breast Cancer Symposium; 2018 Dec 4-8; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2019;79(4 Suppl):Abstract nr P6-13-05.

Resource auto-scaling for SQL-like queries in the cloud based on parallel reinforcement learning

Cloud computing is a technology that provides on-demand services in which the number of assigned resources can be automatically adjusted. A key challenge is how to choose the right number of resources so that the overall monetary cost is minimised. This problem, known as auto-scaling, was addressed in some existing works but most of them are dedicated to web applications. In these applications, it is assumed that the queries are atomic and each of them uses a single resource for a short period of time. However, this assumption cannot be considered for database applications. A query, in this case, contains many dependent and long tasks so several resources are required. We propose in this work an auto-scaling method based on reinforcement learning. The method is coupled with placement-scheduling. In the experimental section, we show the advantage of coupling the auto-scaling to the placement-scheduling by comparing our work to an existing auto-scaling method.

A New Ionogram Automatic Scaling Method

AbstractIonograms generated by ionosondes contain significant information on the ionosphere. The scaling and inversion of the ionograms are important first steps for understanding the ionograms and monitoring the real‐time ionosphere. In this paper, a new algorithm is proposed to automatically scale the ionograms with separated O and X waves based on the Chinese Academy of Sciences Digital Ionosonde. The new autoscaling method is achieved mainly by the image recognition technique, mathematical morphology, and graph theory, and critical parameters are identified based on the ionospheric characteristics. To illustrate the efficiency of this algorithm, 10,560 ionograms from the Chinese Academy of Sciences Digital Ionosonde installed in Huailai (40.40°N, 115.52°E), Ganzi (31.64°N, 99.96°E), Wuhan (30.52°N, 114.31°E), Nanning (22.84°N, 108.33°E), and Xiamen (24.26°N, 118.04°E), in north, south, and central China, are used to test the new algorithm. A good performance in automatic scaling was achieved where the values of foF1 and foF2 both agree with the values of hand‐scaled for 91.4% ionograms. The remaining ionograms are almost unscalable manually due to the trace complicity that requires extensive training experience.

Dynamic Multi-level Auto-scaling Rules for Containerized Applications

Container-based cloud applications require sophisticated auto-scaling methods in order to operate under different workload conditions. The choice of an auto-scaling method may significantly affect important service quality parameters, such as response time and resource utilization. Current container orchestration systems such as Kubernetes and cloud providers such as Amazon EC2 employ auto-scaling rules with static thresholds and rely mainly on infrastructure-related monitoring data, such as CPU and memory utilization. This paper presents a new dynamic multi-level (DM) auto-scaling method with dynamically changing thresholds, which uses not only infrastructure, but also application-level monitoring data. The new method is compared with seven existing auto-scaling methods in different synthetic and real-world workload scenarios. Based on experimental results, all eight auto-scaling methods are compared according to the response time and the number of instantiated containers. The results show that the proposed DM method has better overall performance under varied amount of workloads than the other auto-scaling methods. Due to satisfactory results, the proposed DM method is implemented in the SWITCH software engineering system for time-critical cloud applications.

Kinect센서를 이용한 물체 인식 및 자세 추정을 위한 정확도 개선 방법

This paper presents a method of improving the pose recognition accuracy of objects by using Kinect sensor. First, by using the SURF algorithm, which is one of the most widely used local features point algorithms, we modify inner parameters of the algorithm for efficient object recognition. The proposed method is adjusting the distance between the box filter, modifying Hessian matrix, and eliminating improper key points. In the second, the object orientation is estimated based on the homography. Finally the novel approach of Auto-scaling method is proposed to improve accuracy of object pose estimation. The proposed algorithm is experimentally tested with objects in the plane and its effectiveness is validated.

Elastic Scaling of Cloud Application Performance Based on Western Electric Rules by Injection of Aspect-oriented Code

Abstract The main benefit of cloud computing lies in the elasticity of virtual resources that are provided to end users. Cloud users do not have to pay fixed hardware costs and are charged for consumption of computing resources only. While this might be an improvement for software developers who use the cloud, application users and consumers might rather be interested in paying for application performance than resource consumption. However there is little effort on providing elasticity based on performance goals instead of resource consumption. In this paper an autoscaling method is presented which aims at providing increased application performance as it is demanded by consumers. Elastic scaling is based on “statistical process monitoring and control” and “Western Electric rules”. By demonstrating the architecture of the autoscaling method and providing performance measurements gathered in an OpenStack cloud environment, we show how the injection of aspect-oriented code into cloud applications allows for improving application performance by automatically adapting the underlying virtual environment to pre-defined performance goals. The effectiveness of the autoscaling method is verified by an experiment with a test program which shows that the program executes in only half of the time which is required if no autoscaling capabilities are provided.