A reliable and cost-efficient auto-scaling system for web applications using heterogeneous spot instances

Abstract

Cloud providers sell their idle capacity on markets through an auction-like mechanism to increase their return on investment. The instances sold in this way are called spot instances. In spite that spot instances are usually 90% cheaper than on-demand instances, they can be terminated by provider when their bidding prices are lower than market prices. Thus, they are largely used to provision fault-tolerant applications only. In this paper, we explore how to utilize spot instances to provision web applications, which are usually considered availability-critical. The idea is to take advantage of differences in price among various types of spot instances to reach both high availability and significant cost saving. We first propose a fault-tolerant model for web applications provisioned by spot instances. Based on that, we devise novel cost-efficient auto-scaling polices that comply with the defined fault-tolerant semantics for hourly billed cloud markets. We implemented the proposed model and policies both on a simulation testbed for repeatable validation and Amazon EC2. The experiments on the simulation testbed and EC2 show that the proposed approach can greatly reduce resource cost and still achieve satisfactory Quality of Service (QoS) in terms of response time and availability.

Auto-Scaling Web Applications in Clouds

“…With the obtained model, the auto-scaler can more precisely supervise the resource provisioning process. [Upendra et al 2011], [Gandhi et al 2012], [Fernandez et al 2014], and [Qu et al 2016] employed this approach. The drawback of it is that the profiling needs to be reconducted manually every time the application is updated.…”

“…We proposed an auto-scaler [Qu et al 2016] that uses heterogeneous spot instances to provision web applications. The intention of using heterogeneous VMs in this case is to boost the reliability of clusters based on spot instances to save cost, which is explained in the following section.…”

CMI: An online multi-objective genetic autoscaler for scientific and engineering workflows in cloud infrastructures with unreliable virtual machines

“…It is worth noting that, from the related works found, most of them have been proposed for workflows considering task deadlines or budget constraint but without considering spot instances. Besides, another crucial distinction is that in none of the surveyed works the authors have considered to reduce the impact of OOB errors as part of the optimization process when using spot instances, and only in three works [7,26,27] the authors have proposed to minimize the makespan, rendering difficult their applicability to execute scientific workflows in clouds infrastructures to support rapid domain-specific "in vitro" experimentation. Concretely, in the present paper the objectives are to minimize the makespan, monetary cost and the failure probability when different types of spot instances and bid prices are considered.…”

A heuristic approach to the multicriteria design of IaaS cloud infrastructures for Big Data applications

“…Cost and computing capacity can be intuitively thought of as objectives that cannot be simultaneously optimized, whereas capacity and reliability also collide with each other when a solution maximizing both criteria is sought. In this aspect, the literature is rich in what regards to recent contributions dealing with auto-scaling (Vaquero, Rodero-Merino, & Buyya, 2011;Morais et al, 2013;Ghanbari, Simmons, Litoiu, Barna, & Iszlai, 2012), the comparison or selection between existing cloud solutions (Lenk, Menzel, Lipsky, Tai, & Offermann, 2011;Oki et al, 2017;ur Rehman, Hussain, & Hussain, 2011;Repschlaeger, Wind, Zarnekow, & Turowski, 2013;Do et al, 2016) in different domain applications (Krieger, Torreno, Trelles, & Kranzlmüller, 2017) and solutions (Qu, Calheiros, & Buyya, 2016). Despite this upsurge of research, to the best of our knowledge, there is no prior work dealing with the multicriteria optimality of IaaS infrastructure for Big Data solutions.…”