Non-invasive service dependency finding method

Abstract

The invention discloses a non-invasive service dependency finding method. The method is that a step-by-step calculation method and a method comprehensively considering the use condition similarity and TCP / UDP connection number similarity to measure service distance are designed, aims at focusing special nodes and enabling accurate service clustering; the clustering results of hierarchical clustering algorithm based on the resource use condition and the TCP / UDP connection number are compared to obtain a representation center point which can effectively describe each service class, and then the clustering is performed by the manner of approximating the rest services to the center point; the clustering results are more accurate than that of the simple hierarchical clustering algorithm, and the complexity of the clustering algorithm is reduced. Compared with the traditional KMeans of continuously iterating all services to find out the optimal center group, the method has the advantages that the repeated type filtering manner is carried out, so that the complexity at finding out of the center service can be reduced.

Description

technical field [0001] The invention relates to the field of service dependency discovery in a cloud computing environment, in particular to a non-invasive service dependency discovery method. Background technique [0002] With the rapid development of cloud computing technology, more and more infrastructure, platforms and software are provided to users in the form of services. Moreover, infrastructure service providers, platform service providers, and software service providers are generally not the same. Therefore, infrastructure service providers generally do not understand the specific implementation framework of the software provided by software service providers. Software services in the cloud computing environment mostly realize functional requirements in the form of composite services. How to obtain as many dependencies between services as possible without violating the privacy of software services will help intelligent service scheduling, intelligent data backup,...

AI Technical Summary

Problems solved by technology

Its purpose is to solve the problem of how to discover the dependencies between the services that make up the same application, and provide a method to measure the service distance and automatically complete the service by calculating the similarity between the resource usage and the number of TCP / UDP connections between services. clustering method

Images