Quality of Service (QoS)-Aware Microservices Selection Based on Local Constraints

Abstract

In the world of microservices, to deliver a particular task, there may exist several microservices. Though they are functionally similar, their Quality of Service (QoS), which refers to their non-functional attributes would be different. Quality of Service helps to select services having adequate QoS as per the demands of service consumers so that the composition of microservices can happen within the given QoS constraints. Global QoS based selection which selects microservices at composite service level is associated with inherent exponential time complexity. In this work, a method has been proposed for microservices selection based on local QoS constraints. The given QoS constraints are divided into local constraints corresponding to the number of tasking involved in the process. Then for each task, the services which satisfy the local constraints of the task, are identified as candidate services. Afterwards the service having the highest utility is returned as the best service for that task. The proposed method has been evaluated with a case study and the results are discussed. The case taken for the study is microservices based travel_plan process, consisting of three sequential tasks namely, flight ticket booking, hotel booking and cab booking tasks. The proposed method has found to yield two feasible solutions which guarantee the given global constraints of QoS attributes. Also, in case candidate service is not found for a task, then the method of assigning local constraints is relaxed as follows. For a task, whenever maximum value of a QoS attribute is less than that computed by the proposed method of equal distribution, then the maximum value of the attribute itself will be assigned as local constraint and the remaining QoS attribute is used in the allocation of local constraint of that attribute to other tasks involved in a process.