Bidding Strategies in QoS-Aware Cloud Systems Based on N-Armed Bandit Problems

Abstract

Abstract-In this paper we consider a set of Software as a Service (SaaS) providers, that offer a set of Web services using the Cloud facilities provided by an Infrastructure as a Service (IaaS) provider. We assume that the IaaS provider offers a pay only what you use scheme similar to the Amazon EC2 service, comprising flat, on demand, and spot virtual machine instances. We propose a two-stage provisioning scheme. In the first stage, the SaaS providers determine the number of required flat and on demand instances by means of standard optimization techniques. In the second stage, the SaaS providers compete by bidding for the spot instances which are instantiated using the unused IaaS capacity. We put our focus on the bidding decision process by the SaaS providers, which takes place during the second stage, and apply N-armed bandit problems, in which the player is faced repeatedly with a choice among N different options, and every time he submits his decision evaluating past feedbacks. Through numerical experiments, we analyze proposed strategies under different scenarios and prove the SaaS providers ability to refine their behavior round by round and to determine the best bid so to maximize their revenue and achieve as many spot resources as possible, also addressing the importance of a trade-off between exploration and exploitation, i.e., among greedy and non-greedy actions.

Generalized Nash Equilibria for the Service Provisioning Problem in Multi-Cloud Systems

“…In the first stage, the VM requirements are calculated for each SaaS provider by means of standard optimization techniques, whereas the second stage, modeled as a Stackelberg game, determines the spot price that maximizes the IaaS revenue. The same problem is considered in [2], but the second stage is modeled as an N -armed bandit problem in which every player has to choose among N options taking into account past feedback.…”

“…Constraint (2) ensures that the response time is less or equal to the threshold R k established in the SLA contract, while (3) guarantees that resources are not saturated. Constraint (4) is introduced for fault tolerance reasons and guarantees that the on spot instances are at most a fraction η j < 1 of the total capacity allocated for WS application k at IaaS i.…”

Spot pricing in the Cloud ecosystem: A comparative investigation

“…Furthermore, untruthful bidding and mutual cooperation can cause a cyclical effect and eventually make a market-driven strategy turn to be unprofitable for Cloud providers [KS13,AVCP14]. In fact, the de facto spot pricing in the current Cloud market has been considered to be too primitive to guarantee truthful bidding and fair resource allocation [WRM12].…”

“…-Lyapunov optimization algorithm -Hamilton-Jacobi condition based approach -Mixed-integer nonlinear programming -N-armed bandit -greedy approach [DCL13]). When it comes to solving the revenue maximization problem, Wang et al [WQH + 13] adopted a Lyapunov optimization algorithm with the basic idea of minimizing a bound on the drift-plus-penalty term; Xu and Li [XL13] proposed an optimal solution based on a demand model, the Hamilton-Jacobi conditions, and a standard numerical approach; and Kantere et al [KDF + 11] employed mixed-integer nonlinear programming to optimize the service pricing; and in one of the experimental scenarios, the provider was assumed to set spot prices using an N-armed bandit -greedy approach that repeatedly select the greedy pricing action among a set of N actions [AVCP14].…”

Web Service Negotiation Model Based on Cooperative Game Theory

“…Relative researches could be divided to two sorts: one-to-one static Web Service negotiation [2][3][4][5][6][7][8][9] and complex negotiation [10][11][12][13][14][15][16][17]. However, all of these studies considered one-to-one static negotiation process while the SLA automatic generation algorithm is not achieved.…”