A bartering double auction resource allocation model in cloud environments

Abstract

Resource provisioning and allocation are two essential operations of cloud service providers (CSPs). Due to limited resources, it may sometimes not be cost-effective for CSPs to obtain them from the market. On the other hand, releasing some less profitable resources may violate the service level agreement (SLA). This article presents an auction-inspired model for providing/demanding resources without exchanging money among CSPs. The core idea is that CSPs do not need to pay monetary values. Instead, they exchange free resources through an auction mechanism and are encouraged to cooperate. This model, which we call bartering double auction resource allocation, helps CSPs receive the resources they need in the form of virtual machine instances with different types of functionalities. Under this mechanism, if one party violates the SLA, the other takes back its resources to avoid possible loss. The experimental results using the CloudSim simulator and the statistical analyses show that the proposed mechanism can increase CSPs' social welfare and profitability while significantly reducing SLA violations. Also, the proposed model provides four essential economic properties, that is, individual rationality, incentive compatibility, budget balancing, and economic efficiency.

Entropy‐aware energy‐efficient virtual machine placement in cloud environments using type information

“…Accordingly, some studies focus on reducing the number of contract violations in SLA. [9][10][11] Unfortunately, VMP falls into the category of NP-hard problems. 12 For this reason, classic mathematical solutions such as convex optimization methods fail to solve the problem.…”

“…This, in turn, can minimize SLA violations, and operational costs and hence increase cloud providers' profits. Accordingly, some studies focus on reducing the number of contract violations in SLA 9–11 . Unfortunately, VMP falls into the category of NP‐hard problems 12 …”