Research On Service Selection Strategy For Multi-Tenant Systems Based On Auction

Cloud computing,as a new business model,provides users with ubiquitous,on-demand networkaccess computing resources.With the benefit of convenience,reliability,scalability and low costs,cloud computing has been widely used by enterprises,institutions and so on.Service-oriented system(SOS)is one of the main development paradigms in software engineering.SOS can dynamically compose at runtime and provide sophisticated self-adaptation.Service-oriented system has widely been used in e-business,e-government.Multi tenancy technology is one of the core technologies of realizing business mode of cloud computing.It shares the same stack's hardware and software resources among multiple tenants.By reducing the cost of service delivery and operation cost among tenants to achieve economies of scale.In order to provide a cost effective solution for multiple tenants,SOS must achieve multi-tenancy.Quality-aware service selection for service-oriented system is a key problem.Service selection for a single-tenant SOS is a NP problem.To compose a multi-tenant SOS,the system developer needs to select appropriate services that collectively fulfils all tenants' QoS constraints(e.g.,throughput,response time,reliability,reputation and availability),and achieves the SOS owner's optimisation goal(e.g.,minimum system cost or maximum system utility).Thus,the service selection problem is more complicated for multi-tenant SOS.In the past decade,many approaches for service selection have been proposed to compose single tenant SOS and multi-tenant SOS.However,none of the existing approaches have systematically considered three critical issues to increase success rates of service selection:1)multi-tenancy,i.e.,the ability for an SOS to serve multiple tenants simultaneously;2)provider competition,the competition between service providers has not been fully explored;and 3)service complementarity,i.e.,the complementarity between the quality of multiple services.Auction has been proven to be an effective and efficient one-to-many negotiation method,and increasing services selection efficiency and social benefits,especially when the items to be allocated exhibit complementarities.Auction can capture the preferences of both buyers and sellers and ensured their satisfaction.Therefore,we propose auction to support effective and efficient service selection for multi-tenant SOS.The major work of this paper is as follows:1.There are a large number of tenants in cloud computing,and services that tenants need is different.Furthermore,tenants usually have differentiated multi-dimensional quality constraints for services.In order to offer cost-effectively solutions to multiple clients,an SOS must achieve multi-tenancy,i.e.,the ability to satisfy multiple clients simultaneously.The service level approach for building SOS at the second maturity level demonstrates the best performance in finding a solution.Furthermore,the data isolation enabled by the second maturity level offers the best data security and privacy for tenants.During service selection,the system developer should negotiate service level agreements(SLAs)with service providers and then select a solution that achieves the system developer's optimization goal while fulfiling all tenants' quality constraints.This exacerbates the competition among service providers.Negotiation is a strong incentive for service providers that can provide multiple types of services to propose competitive QoS offers and upgrade their QoS offers in order to win more service contracts.However,traditional service selection approaches for multi-tenant SOS fail to fully explore the competition among service providers to increase the possibility of finding an optimal solution.Therefore,we focus on the second maturity level of multi-tenancy,and propose a novel iterative auction based service selection for multiple tenants systems.2.Service providers' profits increase when the numbers of service contracts that they win increase.Considering the complementarity between services,we allows service providers to bid for combinations of services in a structured manner.This feature allows service providers to utilise their QoS capacities for complementary services.They can apply discounts or premiums to their QoS offers for combinations of services.Bidders with the ability to offer multiple services will be able to gain advantages over other service providers during the SLA negotiation,as for they can provide better quality of service at lower price.Existing service selection approaches fail to fully consider the competition among service providers and the complementarity between services to increase the possibility of finding an optimal solution.Therefore,we focus on the second maturity level of multi-tenancy,and propose a novel combinatorial auction-based service selection for multi-tenant systems.3.We consider the competition among service providers and the complementarity between services to increase the success rate of service selection for multi-tenant SOS.There are four multitenancy maturity levels.We propose combinatorial auction-based service selection approaches for composing SOS that achieve the third and the fourth multitenancy maturity levels respectively.This paper has analyzed four multitenancy maturity levels,and proposes auction-based service selection approaches that optimise multi-tenant SOS.Experimental results show that our method can significantly improve the success rate of finding a solution and system optimality.Meanwhile,its efficiency,measured by the number of auction rounds and computation time,is demonstrated to be satisfactory in scenarios on different scales.