A Review of Auto-scaling Techniques for Elastic Applications in Cloud Environments

Abstract

Cloud computing environments allow customers to dynamically scale their applications. The key problem is how to lease the right amount of resources, on a pay-as-you-go basis. Application redimensioning can be implemented effortlessly, adapting the resources assigned to the application to the incoming user demand. However, the identification of the right amount of resources to lease in order to meet the required Service Level Agreement, while keeping the overall cost low, is not an easy task. Many techniques have been proposed for automating application scaling. We propose a classification of these techniques into five main categories: static thresholdbased rules, control theory, reinforcement learning, queuing theory and time series analysis. Then we use this classification to carry out a literature review of proposals for auto-scaling in the cloud.

A control theoretical view of cloud elasticity: taxonomy, survey and challenges

“…A comprehensive survey on cloud elasticity is carried out in [7], where the authors classified the overall elasticity literature based on the underlying implementation techniques. Galante and De Bona [1] classified them into infrastructure and application level, and a taxonomy consisting of features like scope, purpose, decision-making mechanism, action type and evaluation is proposed in [8].…”

“…All the survey papers reviewed are very innovative and mostly overlapped regarding the essential elasticity features, e.g., elasticity type (Reactive/Proactive), trigger (Horizontal/Vertical), scope [Cloud provider (CP)/Service provider (SP)], etc. However, their scope is wide, i.e., overall cloud elasticity research and apart from [7], they lack details on the underlying implementation techniques of the proposed solutions. Majority of the existing such review papers utilise the various cloud elasticity features to classify the overall cloud elasticity literature.…”

“…Majority of the existing such review papers utilise the various cloud elasticity features to classify the overall cloud elasticity literature. In such a classification, the underlying implementation technique is consider as one attribute, (e.g., in [7,9]) and therefore no classification of the existing literature is performed using the attributes of a particular implementation technique. This result in the lack of conducting an exhaustive review of the proposals of each implementation technique.…”

“…There are also several survey papers available that provide a concise review of different aspects of cloud elasticity. Lorido-Botran et al [7] classified the overall elasticity proposals based on the underlying implementation techniques, whereas Naskos et al [8] distributed the literature based on the decision-making mechanism, and Coutinho et al [9] on the other hand provided a systematic review of utilised performance metrics, measurements tools and evaluation. The scopes of all these papers have been broad where they mainly focused on the high-level view of overall elasticity research rather than the specific details on one implementation technique.…”

“…This paper however, does not cover the following aspects of elasticity domain including the commonly used experimental platforms, real workloads' data, monitoring tools, application benchmarks. The key reason behind is that these important aspects are already fully covered in various related survey papers such as [7,10]. This paper will help researchers of the target area to understand the various related concepts of cloud elasticity including control theoretic approaches, different possible types of control solutions, how these are used to implement cloud elasticity, their pros and cons and the open research challenges.…”