The Huaqiao University in Quanzhou (China) organized the conference: The 2nd International Conference on Modern Management based on Big Data (MMBD2021) took place remotely (using the MS Teams platform) on November 8-11, 2021. The paper entitled: The Support System for Anomaly Detection with Application in Mainframe Management Process was presented by MSc. Alicja Gerka.
The aim of this presentation was to present a paper that has already been published as a post-conference material in the series of IOS Press: Frontiers in Artificial Intelligence and Applications, in volume 341: Modern Management based on Big Data II and Machine Learning and Intelligent Systems III, on pages 96- 103:
https://ebooks.iospress.nl/volumearticle/57956
Strzałka, D., Gerka, A., Kowal, B., Kuraś, P., Leopold, G., Lewicz, M., & Jaworski, D. (2021). The Support System for Anomaly Detection with Application in Mainframe Management Process. In Modern Management based on Big Data II and Machine Learning and Intelligent Systems III (pp. 96-103). IOS Press.
As part of cooperation with Z-RAYS company, the Department of Complex Systems employees: D. Strzałka, B. Kowal, A.Gerka. P. Kuraś, M. Ćmil took part in the preparation of a solution using machine learning to build an anomaly detection support system in mainframe machines.
The cooperation included the annual involvement of the team and the implementation of the following activities: selection of a subset of the derived metrics, development of sources and development of effective algorithms for their processing; analysis and development of intelligent event correlation algorithms; integration of the developed algorithms and algorithms embedded in the ELD/Dynatrace tools into the Z-RAYS software; development and integration of data collection methods and implementation of Z-RAYS on the SaaS platform.
The process of managing and administering mainframe machines requires not only specialized expert knowledge based on many years of experience, but also appropriate tools provided by the machine performance management system, including Resource Measurement Facility (RMF). The aim of the article was to present the initial results of building the Z-RAYS system based on machine learning (ML) techniques. It automatically detects anomalies and generates early warnings of some errors that may appear on the mainframe to support the mainframe management process. The presented results are based on extensive simulations based on the IBM emulator. We focus on determining the degree of metric variability, the degree of data repeatability in metrics, some approaches to detect metric anomalies, and solutions to detect event correlation in metrics.
