Research And Application Of Band-area Business Container Model

In the cloud environment,with the rapid increase in the number of people and application services and the close integration of people and application services,software development patterns are also constantly evolving.Cloud service providers must consider this problem to integrate users and existing enterprise business systems into a complete enterprise application system.In the process of application system integration in cloud platforms,the adoption of application container virtualization methods changes the manner in which applications are built,deployed,and migrated.Second,the combination of container virtualization technology and microservice frameworks affects the development,deployment,and scheduling processes of applications.Research on building cloud applications based on a combination of application container virtualization technology and microservice frameworks is an important research topic in the cloud.However,most application containers focus on the interface between the application and host machine,as well as computing resource management,rather than business processing,and lack a collaborative mechanism between application containers.To solve the problem of collaboration between application containers and achieve resource management,this paper studies a resource management framework that combines a band-area business container based on application container virtualization technology with tool services based on microservice frameworks.The main research work and innovative points are as follows.(1)To solve the problem of difficult resource management,a new resource management model,the Band-area Business Container(BAC),was studied.BAC includes objects such as users,tool services,documentation,messages,and a set of related operational rules.To facilitate the interaction of business data between tool services,three invocation relationships between services are proposed: pipeline,production-consumer,and random handshake relationships.This enables users to combine tool services into Service Function Chains(SFCs)in BAC through the invocation relationships between services to construct new sub-business application systems.To meet the requirements of building more granular business applications through composition between multiple application containers that are business driven but independent of specific businesses,two composite patterns between BACs are proposed:sequential and parallel.In response to the lack of business collaboration mechanisms between application containers,five types of association relationships between BACs have been established: inheritance,inclusion,I/O,homology,and companion relationships.To improve the reuse of BACs,a Band-area Business Chain(BBC)was established based on the association relationship between BACs,providing convenience for users in building business application systems by reusing the BBC.A BAC system architecture has been proposed,which provides a theoretical basis for users to combine BACs into coarse-grained business application systems through the association relationships between BACs.To provide a running support environment for business application systems built based on a combination of BAC and tool services,the BAC system framework and Baa S platform,which can provide users with favorable application services and operating environments,were studied.(2)The formal semantic method is used to formally describe tool services,invocation mechanism between services,BAC model,BAC operation rules,composite pattern and association relationship between BACs,and BAC architecture,and to establish and form a formal description framework and basic theoretical framework based on the BAC model and association relationship.The assembly and reuse process based on BAC,tool services,and corresponding collaborative mechanisms was established through formal methods,and the feasibility of constructing a new business application system through the combination of BACs was formally verified.This provides a formal semantic framework and a standardized definition for implementing business application system integration based on BAC models and association relationships.(3)To reduce the response time of applications and improve the resource utilization of cloud data centers(CDC),objective functions such as service transmission cost,resource load balancing of the CDC,and container aggregation value were established.A service container deployment strategy based on an accelerated particle optimization algorithm(APSO-TSDS)is proposed.This strategy is to cluster the service containers in SFC as much as possible to the same physical node or the same CDC,and shorten the network distance between services through container aggregation to reduce the data transmission cost of services and improve the resource utilization of the CDC.The experimental results show that compared with existing deployment strategies,the APSO-TSDS strategy has significantly improved performance indicators,such as container aggregation value,service transmission cost,and resource utilization of the CDC,reducing service transmission cost,and improving resource utilization of the CDC.(4)Based on the application requirements of deploying services of different business types(such as computationally intensive and storage-intensive)to the corresponding types of server nodes,three objective functions were constructed by analyzing the similarity between service containers and server nodes: compatibility between containers and server nodes,load balancing of clusters,and service execution reliability.A container deployment strategy for different types of services(AF-CSDS)was proposed.This strategy finds suitable and reliable server nodes for a service container under the constraint of balancing the resource utilization of single and multiple nodes.The experimental results show that the AF-CSDS deployment strategy significantly improves the resource utilization of computing and storage server nodes;In addition,while ensuring system performance and efficiency,the load between each type of server node is balanced.(5)To solve large-scale concurrent tasks and find a suitable container and node for task execution,objective functions such as the task processing time cost,energy consumption,and failure rate were established,and the resource utilization of containers and nodes was also considered.A multi-objective optimization algorithm for artificial fish schools based on multi-layer container task scheduling(AF-MLCS algorithm)is proposed.This algorithm determines a suitable container for each task by simultaneously optimizing multiple objectives and allocating the selected container to the server with the optimal resource utilization standard deviation of the node.The experimental results show that,compared with existing scheduling algorithms(APSO-ECS,GA-MOCS,Binpack algorithm),this algorithm improves the task processing time by 17.29%～74.21%,energy consumption by 25.28%～73.1%,and failure rate by 24.19%～43.66% while balancing the resource utilization of the CDC.Maximizing the reduction of task processing time and energy consumption and improving the stability of task execution.