The Research On Resource Cache Mechanism In Fog Computing Environment

In today's society,cloud computing technology has been widely used,but also exposed a series of problems.A prominent problem is the decline in quality of service caused by large time delay.First of all,computation and storage of cloud computing are centralized in the cloud data center.The data transmission distance between terminal devices at the edge of the network with the cloud data center is long,so as to bring time delay.Secondly,the transmission and execution of a large number of tasks also bring huge load pressure of network bandwidth and cloud data center,resulting in an increase of time delay.In 2011,Cisco proposed the concept of fog computing,which extends the computation and storage of cloud computing to the edge of the network.Devices at the edge of the network perform part of the computation and storage tasks,so as to shorten the transmission distance and reduce the load pressure of network bandwidth and cloud data center.In addition,cache is also an important way to reduce time delay in the network.In this paper,aiming at the problem of cloud computing,combining fog computing with cache,we propose a resource cache mechanism in fog computing environment.The research work mainly includes four aspects.Firstly,the resources cache system in the fog computing environment is designed.The system is divided into two parts:the management end and the fog node cluster.The fog node cluster is composed of a number of terminal devices as fog nodes and is responsible for caching resources.The management end is the control core of the entire cache system,including a metadata management module,a load evaluation module,a cache replacement module,a cache validity check module and a cache storage module.The load evaluation module obtains and ranks the load weights of the fog nodes according to the load balancing algorithm proposed in this paper.The metadata management module mainly manages the metadata of the resources cached in the fog node cluster.The cache validity check module is mainly responsible for ensuring the freshness of the cached resources.The cache replacement module is mainly responsible for the replacement by the optimal cache value replacement algorithm proposed in this paper when the physical space is insufficient.The cache storage module is responsible for caching resources in a distributed multi-replicas form in the fog node cluster.Secondly,in order to ensure robustness of cache system and reduce the time delay.this paper proposes a distributed multi-replicas storage strategy,which integrates power supply,switch and remaining physical capacity.In order to balance the fog node cluster load to reduce the time delay,this paper proposes a load balancing algorithm,which integrates CPU utilization,memory utilization,IO utilization,bandwidth utilization and hardware metrics,to select fog node to response request.Afterwards,this paper proposes an optimal cache value replacement algorithm.The algorithm idea is to calculate the value of the resource,according to access frequency of resource,size of resource and remaining validity period ratio of resource,when the physical space is insufficient.Then,according to value,resources are selected for replacement to ensure the sum of value of cached resources is maximum after the replacement.Finally,through the concurrent access test of network resources of different size,thispaper verifies the effectiveness of the storage strategy and the load balancing algorithm in reducing the time delay and balancing load.Through time delay and byte hit rate test by using the web log,this paper verifies the feasibility of the cache replacement algorithm.Experiments show that the resource cache mechanism in the fog computing environment proposed in this paper can effectively reduces the time delay and improves quality of service.