A method for Industrial heterogeneous big data service application integration

In recent years, data analysis services have gradually used in industrial enterprises. To satisfy complex and changing business increase, more and more data analysis services are needed in enterprises. There are significant differences in different data analysis service application, that make it is difficult to reuse analysis results and interconnecting services. Therefore, how to integrate, manage and use these services in a unified manner is a key issue for industrial enterprises. To solve these problems, we propose a heterogeneous big data service integration framework. A existing big data analysis service is divided into three parts: data input, data analysis and processing, and service results, which respectively solve the data input integration problem, data processing and analysis integration problem, and output result integration problem. In addition, to address the service scheduling problem under limited enterprise resources, we propose a service scheduling method based on density peak clustering. Based on the clustering results of historical service requests, the time period is dynamically divided into multiple fine-grained time intervals to accurately mine service request patterns. Experimental results show that our method can set the service concurrency threshold more accurately and maximize the number of service executions under limited resource conditions.