Research On Microservice Migration In Mobile Edge Computing

Mobile Edge Computing(MEC)offloads the service resources originally deployed in the cloud data center to the network edge closer to the user.Microservice architecture decomposes a traditional monolithic application into a set of independent lightweight services that can be executed individually.With microservices,service providers can efficiently deploy innovative applications in MEC.In order to solve the contradiction between user mobility and limited coverage of the edge clouds,it is necessary to migrate microservices between edge clouds.However,most of the research on MEC service migration uses a coarse-grained method to migrate services in the form of the monolithic application.Applying this method to the multi-instance microservice migration will lead to some problems such as lack of dependency,high delay,and high overhead.Therefore,this paper focuses on the migration of multi-instance microservices in MEC and aims to achieve(1)high migration success rate,(2)low migration delay,(3)low communication delay between microservices after migration.The main contents are as follows:1.Aiming at the fact that the existing monolithic architecture migration algorithm cannot solve the problem of the large numbers and the strong dependence of microservices,this paper proposes a low latency multi-instance microservice migration algorithm.Firstly,a delay factor integrating migration delay and runtime delay is designed to measure the fitness of the destination nodes.Secondly,the node reordering algorithm is used to reschedule the migrated nodes for avoiding introducing additional communication delay.Simulation results show that the Low Latency Migration(LLM)algorithm achieves a higher delay benefit and migration success rate than Bandwidth Original Migration(BOM)and Aggregation Migration(AM)algorithm.2.Aiming to further optimize the migration delay and improve the availability of migration,this paper proposes a low latency migration system based on the shared microservice mechanism.The main idea of the shared microservice mechanism is to identify and share redundant microservices,so as to reduce the number of microservices that need to migrate.Firstly,a shared resource analyzer and a runtime monitor are designed to collect and monitor shared resources.Secondly,a selection strategy based on the shared migration factor is proposed,which enables microservices to select the best target node with or without sharing.Experimental results show that the proposed system reduces the migration delay of microservices and avoids resource conflicts between different applications on the sharing microservices.3.When there are fewer microservices that can be shared,the benefit of the shared microservice migration system is low.To solve this problem,this paper proposes a migration strategy based on the microservice caching mechanism.The main idea is to obtain the hot microservices and deploy them in the cache space in advance,so as to improve the hit rate of shared microservice.Firstly,for the sake of meeting the requirements of multiple applications using the same cache object at the same time,a microservice cache update strategy with high popularity and low storage priority is proposed.Secondly,a migration strategy based on the cache migration factor is proposed,in which the server nodes with corresponding microservice cache objects are selected as far as possible.Experimental results show that the proposed strategy significantly improves the hit rate of microservice cache resources and reduces the migration delay.