Ex-post efficient resource allocation for Self-organizing Cloud

Abstract

Self-organizing Cloud (SoC) is a very scalable model, which aims to make full use of the geographically distributed computers to provide powerful aggregated computing ability. The resources are provisioned elastically according to user's specific demand, by leveraging virtual machine (VM) resource isolation technology. The resource allocation atop SoC is very challenging in that it is not only an optimization problem over multi-dimensional divisible resources but closely related to economic analysis for coordinating the social competitions. Existing economy-based approaches, however, are commonly ex-ante efficient, such that the participants' decisions may rely on their subjective judgments about their competitors' decisions, leading to the unguaranteed payoffs. Instead, we propose an ex-post efficient resource allocation scheme, which owns three features. (1) Expost win-win effect: under such a scheme, each participant (including consumers and suppliers) will always feel satisfied with their ex-post payoffs. (2) Ex-post incentive compatibility: we can theoretically prove each rational consumer and contributor will always get the optimal payoff if and only if their resource demands and expected prices are truthfully declared. (3) Ex-post maximized efficiency: truly advanced (or more powerful) resources will be allocated and consumed with higher priority, such that the whole system will stay very efficient with maximized resource utilization. Via simulation, our approach can significantly improve the resource contributor's payoffs, also with a high quality of service throughout the global system (including user task's execution efficiency, overall utility, payment level, and fairness of treatment, etc.).

Tender of computational works in heterogeneous distributed environment

“…A combinatorial Vickrey auction provides an opportunity to make offers (bids) for combinations of such lots [24]. An extensive bibliography confirms the effectiveness of applying various forms of such an auction in relation to distributed computing [25][26][27][28][29][30][31][32][33].…”

Impact of Resource Management and Scalability on performance of Cloud Applications - A Survey

“…Sheng Di et al, proposed ex-post efficient resource allocation for self-organizing cloud. [17] Scalability heavily depends on the support of efficient network communications in virtualized environs. Roberto R. E. et al, analyzed the main bottlenecks of performance in HPC application scalability on Amazon EC2 and as a result it was echoed that network communication is the key performance logjam for scalability.…”

Resource discovery for distributed computing systems: A comprehensive survey

“…Cloud computing [1,131,132] provides infrastructure (Infrastructure as a Service or IaaS), platform (Platform as a Service or PaaS) and software (Software as a Service or SaaS) as services to the end users. It helps the users to create and manage their customized hardware and software, while it reduces the cost and the complexity of maintaining the required hardware/software for different user's applications.…”