Multi-tenant Resource Scheduling Method Based On The Quality Of Service

With the continuous development of cloud computing technology, software as aservice SaaS (Software as a Service) as a new software model is widely used.SaaSservice model at the top level of the cloud services, directly interact with the user, thesharing of resources between the users is the fundamental starting point for SaaSservice model, multi-tenant technology as the core of SaaS, the key technologies toachieve this goal.Multi-tenant technology is the same application instance, leased to multipletenants lease manner, sharing of resources between all tenants. Multi-tenanttechnology is based on the lease, as a starting point to integrate resources andimprove resource utilization. Multi-tenant technology has greatly improved theutilization of resources, shared among multiple tenants, an application example, toimplement the technology has brought a very big problem: first, the use of resourcesbetween the tenants interfere with each other, reduce the efficiency of resource use,quality of service of tenants is difficult to be guaranteed; Secondly, when the resourcepool reaches a certain size, the efficiency of the multi-tenant resource schedulingalgorithm can not meet the actual needs.In order to achieve a good multi-tenant technology to solve the above technicalproblems, from the forms of organization and user level of the resource pool isdivided into two aspects, a multi-level resource pool based on the QoS level oftenants by hierarchical resource scheduling strategy. The main work of this paper is asfollows:(1) In order to improve the efficiency of resource scheduling, the establishmentof three resource pool structure.First underlying SaaS resource pools by propertyclassification, the formation of the four classification resource pool;In each categoryresource pool, the establishment of two resource QoS value keyword index table toform the index resource pool.The overall three resource pool structure, the resourceson the basis of classification according to the QoS property values ordered by theindex structure to speed up the scheduling of resources. (2) In order to ensure the quality of service requirements of different tenants, thetenants grade division to determine the order of execution of the request;according tothe type of request needs resources, divided into four categories of sub-request andclassification of resource pool corresponding; QoS levels according to the sub-requestdifferent scheduling strategies: sequential scheduling, forward half of the scheduling,moved half scheduling. Three ways to find the two indexes resources table fromdifferent locations in different directions to reduce mutual interference and a half wayto speed up resource scheduling.(3) To demonstrate the effectiveness of the method to build a simulationenvironment of the proposed method, experimental results show that the way toachieve quality of service requirements of the tenants, to ensure the isolation ofresource scheduling, and hashigher efficiency of the implementation.