| Cloud computing has been widely used in the field,but with the development of Internet of Things technology,cloud computing is facing many problems that need to be solved urgently.Because of the expensive construction cost,cloud computing can not achieve large-scale deployment,and cannot process data of a large number of terminal devices in the Internet of Things in time,and cannot meet the application requirements of delay sensitive and location-aware.Cisco predicts that the number of connected devices worldwide will reach 50 billion by 2020,and with the rapid growth of Internet of Things devices,massive amounts of data will be transferred to the data center for processing,and by the end of 2020,the global data centers will have 15.3 ZB of IP traffic per year.If the Internet of Things continues to use the current cloud computing paradigm to handle large amounts of devices and data,it will result in high latency and network congestion.Based on these issues,Cisco proposed the concept of fog computing in 2012 and defined and elaborated on fog computing in 2013.To support real-time applications and reduce the burden on data center networks,data generated at the edge of the network is cached and processed on the fog side in a timely manner,rather than being stored almost entirely in the cloud data center.Deploying computing devices at the edge of the network is an important means of addressing cloud computing transmission latency in the service process.Fog computing is proposed as a remedy for the shortcomings of cloud computing.There are many similarities between fog computing and cloud computing.For example,they are used by virtualization techniques to abstract the underlying physical resources to serve the user.This paper discusses the application of container virtualization technology in fog computing and the management of virtualized resources.Based on in-depth study and learning of virtualization technology and task scheduling and resource management methods,we combine the relevant characteristics of container virtualization technology to optimize management for computing resources,storage resources and network resources.Firstly,aiming at the problems of high deployment cost and difficult maintenance in fog computing,a system architecture of multi cloud and multi fog collaboration based on container is proposed.Secondly,a task scheduling algorithm based on dynamic threshold and resource redistribution mechanism based on dynamic threshold are proposed to improve the computing resource utilization of nodes in view of the task scheduling problem of fog computing in the smart factory environment.Thirdly,according to the hierarchical structure of image file of container virtualization technology,an optimized image placement strategy is proposed to improve the utilization of storage resources.Finally,in response to the problem of excessive delay in service migration caused by frequent movement of terminal devices in urban environment,a mobile-aware service migration method is proposed,which reduces data forwarding delay and improves network resource utilization.Summarize the full text,there are mainly the following research results:1.As the construction cost of data centers decreases and the number of private clouds increases,the traditional single-cloud and multi-fog architectures are gradually showing weaknesses in maintenance costs and service diversity.We established a system architecture of multi cloud and multi fog collaboration based on container.It defines two types of service types: temporary service and long-term service in fog computing,as well as infrastructure providers,service developers and end users.At the same time,a detailed description is given to the three forms of the service process,request-service-container.Through software-defined services and cloud agent methods,service developers are provided with a service publishing platform,which makes the fog computing service more flexible,and solves the problem of multicloud and multi-fog nodes cooperating,thereby improving resource utilization and reducing node construction and maintenance costs.2.Although a large number of studies have applied containers to cloud services,it is urgent to solve the problem of delaying service delay and limited resources in fog computing.How to combine the characteristics of containers to efficiently utilize the limited resources of fog nodes to provide faster service.problem.This paper deeply analyzes the image file storage mechanism of the container and builds an image file storage model based on fog computing.A two-stage storage strategy for image files is designed for the hierarchical structure of image files.Firstly,in the initialization phase,the maximum transfer amount of the image file in the initialization phase is obtained by hierarchical deconstruction of the image file,and the optimal image file combination satisfying the current node storage space is selected by the combination optimization method.Then,in the runtime phase,the frequency of use of the image file is analyzed for all requests of the node,and the image file of the local image library is periodically adjusted and optimized.for image files through optimized placement during the initialization phase and optimized updates at runtime can effectively reduce download time of images and service latency.3.At present,most task scheduling algorithms for fog computing are based on virtual machines.With the rapid development of container technology,containers are gradually replacing virtual machine technology into the mainstream virtualization technology in fog computing.In the smart factory environment,the task delay sensitivity is high and the computation-intensive tasks are many.We construct a task scheduling model for fog computing in a smart factory environment.The container is used as the task unit to provide computing resources to the terminal equipment,and the fog-based task scheduling model based on smart factory is constructed.Then based on the characteristics of the container,the task scheduling process is subdivided into three phases: the request evaluation phase,the task scheduling phase,and the resource redistribution phase.When the request reaches the node,it first determines whether the current node satisfies the delay constraint of the request.For the request that meets the delay constraint,it enters the task scheduling phase.In the task scheduling phase,a cloud-based task scheduling algorithm is proposed to schedule tasks.Finally,in the resource redistribution phase,an optimized resource redistribution scheme is proposed by using the resource dynamic adjustment mechanism of the container.The resource redistribution method adjusts resource quotas for all tasks of the current node according to the task quantity of the current node,thereby improving resource utilization of the fog node and reducing service delay.4.The object of the fog computing in the urban environment is characterized by high mobility.How to provide low-latency services in the case of fast movement of service objects is one of the problems that fog computing needs to solve.Real-time service migration is one way to solve this problem.However,the container has process nesting and strong association with the image file,resulting in a lack of real-time migration.This paper studies the service migration of fog nodes in urban environments.In the urban environment,the service-intensive area also increases the service delay of the node,thereby reducing the quality of service.In order to reduce the service delay of nodes,this paper proposes a mobile migration-based service migration mechanism.The mobile device-based service migration mechanism triggers the service migration according to the service density of the current node,and selects the service set that is most suitable for migration in the current node with the minimum service delay as the optimization target.Finally,the traffic direction and the migration cost of the device to which the service belongs are selected according to the service direction.The destination node is migrated.The mobile sensing-based service migration mechanism can effectively reduce the waiting delay and migration delay in the service process,and further optimize the service quality of the fog node.This paper deeply studies the characteristics of container virtualization technology,and applies the container virtualization technology in fog computing,combined with the actual scenarios of smart factories and smart cities,the key indicators of service quality such as service delay,resource utilization rate and concurrent number are taken as the optimization objectives,a resource model of fog node based on container is constructed,In addition,the computing resources,storage resources and network resources in fog computing are optimized and scheduled,which improves the resource utilization efficiency of fog nodes and the user service quality of fog computing. |
PDF Full Text Request