Statistical compaction of a monitoring cloud cluster resource when processing streaming services

Abstract

The scientific problem of developing a method of statistical compaction of the necessary cloud resource in the processing of streaming services in the interests of monitoring systems for various purposes is formulated and solved. The method allows you to determine the number of virtual connections for different services that can be processed in a cloud cluster with a given quality, taking into account, on the one hand, the flow of requests from sources of different services for the provision of virtual connections, and, on the other hand, the needs of these same services transfer rates, taking into Analytical relations for calculation of necessary productivity of a cloud cluster at the set value of probability of loss of a micropackage of streams of DStream are received.