Research On Key Technology Of Shared Resource Management Of Petroleum Exploration Software Platform With Multiple Tenants

Petroleum exploration software is an important and indispensable technological means for supporting oil and gas exploration.However its exclusive resource utilization mode leads to in a certain extent the situation of resource waste and shortage.The rise of cloud computing provides a possibility for shared resource application.As an example,single case multi-tenant mode is a way to use the same case to provide service for different tenants,which becomes a popular service in reducing cost on scale effect.World famous oil companies and petroleum service companies tend to develop their exploration software in cloud computing.So far a software application service mode is provided for their clients which includes concentration and integration of information resources,centralized management of server and application software,etc.Technically some policies like data split,multiple stage migration and qualitative prediction of workload are used.Domestic oil exploration software application on cloud platform has not set up a system analysis method to make allocation strategy and effective migration control to solve the problem.They use the way of platform extending or hanging a low level operation to meet the enlarging needs of high-level tenants.Basically it is rare for them to consider migration of computing resources and data resources.When petroleum exploration software application platform is set up in the cloud based on multi-tenancy mode,different applications oil branch business need to run on the same platform.The platform assigns appropriate resources for different users and completes user submitted business request,and at the same time,meets other non-functional requirements put forward by different users.The contradiction between specific cloud resource scale and tenant resource needs becomes an outstanding problems for multi-tenant application on cloud computing platform,i.e.multi-tenant service level agreement(SLA).Generally speaking,petroleum exploration software application in cloud needs to urgently trade such difficult problems like placement of application software and huge amounts of data,multi-tenant SLA satisfaction and tenant migration with workload variation sensation.This paper focuses its research on the above scenario and requirement including shared resource distribution,workload forecasting,dynamic resource adjustment etc.A shared resource management framework is created for multi-tenant petroleum exploration software application platform.A two phases and dynamic cloud resource allocation strategy is put forward.A SLA fuzzy prediction principle of multi-tenant petroleum exploration software application is formed.Tenant migration strategy based on workload awareness is designed.Shared resource management prototype of multi-tenant petroleum exploration software application platform is developed.Tests show the correctness of research ideas and the feasibility of the results.The main research results include:(1)Shared resource management framework of multi-tenant petroleum exploration software application platformAn overall platform framework is designed for exploration software application sharing resource management.Some challenges are analyzed about the management framework.Workflow of multi-tenant petroleum exploration software application sharing resource management framework is formulated.The shared resource management framework includes mainly management of strategy,initial placement of sharing resources,monitoring,forecasting,migration,etc.All the parts work together to thoroughly implement management of big data oriented multi-tenant shared resource.Tests show the established shared resource management framework in this paper is in conformity with actual application requirement.Effective management of shared resources can be implemented.(2)Multi-tenant data and application placement strategy based on user experienceA two-stage dynamic placement strategy and tenant application aiming at cloud sharing resources and data are put forward for petroleum exploration industry application.The strategy can be realized the initial placement of tenant application and data according to the tenant SLA level and cloud resource utilization.It turns the placement problem of tenant data into two consecutive phase decision problems.The candidate nodes filtering and choice of final placement node are implemented through programs.A DF-based multi-level feedback queue scheduling algorithm is put forward to realize application node perception to tenants SLA.The strategy verified by experiment can reduce complexity of placement problem.It can also ensure high resource utilization while meantime satisfies tenant good user experience.(3)SLA fuzzy forecast for multi-tenant petroleum exploration software applicationsA principle of multi-tenant petroleum exploration software application platform sharing resource management is established based on the grey theory,i.e.multi-tenant application execution fuzzy prediction.Reasonable definition of the forecasting problems is given.Prediction mechanism is formed based on the grey theory.Local prediction and comprehensive prediction model are established.An experimental environment is established and experimental results are analyzed which show the established SLA fuzzy prediction is feasible for multi-tenant oil exploration software application.(4)Tenants migration strategy based on workload awarenessDefinitions of multi-tenant migration,migration rule and migration algorithm are presented.Target node determination is defined.Screening strategy is established.Based on particle swarm optimization an algorithm of particle load sensing mode is proposed.A shared resource oriented dynamic adjustment mechanism is established.Appropriate migration tenants are chosen before load overload of placing node,resources migration strategy is formulated in the dynamic migration process.The strategy is tested under the condition of different threshold value and number of tenant.The results show that tenants migration strategies based on workload sensitiveness of this paper are effective.This paper creates the shared resource management framework and allocation strategy of dynamic two-phase of the cloud resource.It has effective to promote petroleum exploration software application in cloud computing mode.Using the framework proposed in this paper can reduce resource conflict for multi-tenant applications.The SLA fuzzy forecast for multi-tenant petroleum exploration software applications and multi-tenant migration strategy by workload awareness given in the paper are feasible.It has realized reasonable and fast transfer of multi-tenants which also ensures multi-tenant SLA satisfaction.Prototypes and their tests in this paper show that the results have practical significance for multi-tenant petroleum exploration software application platform.