The Support System for Anomaly
Detection with Application in Mainframe
Management Process
Dominik STRZALKA
a
, Alicja GERKA
a, 1
, Bartosz KOWAL
a
, Pawel KURAŚ
a
,
Grzegorz LEOPOLD
b
, Michal LEWICZ
b
and Dawid JAWORSKI
c
a
Department of Complex Systems, Rzeszów University of Technology,
Al. Powstańców Warszawy 12, 35-959 Rzeszów, POLAND
b
Z-RAYS, Plac Andersa 7, 61-894 Poznań, POLAND
c
Department of Mathematical Modelling, Rzeszów University of Technology,
Al. Powstańców Warszawy 12, 35-959 Rzeszów, POLAND
Abstract. The process of mainframe machines managing and administration
requires not only specialized expert knowledge based on many years of experience
but also on appropriate tools provided by a machine performance management
system, e.g. the Resource Measurement Facility (RMF). The aim of this paper is to
show some preliminary results of Z-RAYS system construction that is built basing
on machine learning (ML) techniques. It allows automatic detection of anomalies
and generation of early warnings about some errors that can appear in the
mainframe to support mainframe management process. Presented results are based
on extensive simulations that were done basing on the IBM emulator. We focus on
determining the degree of the metrics variability, the degree of the data
repeatability in metrics, some approaches in metrics anomaly detection and
solutions for event correlation detection in metrics.
Keywords. Mainframe, anomaly detection, support system, machine learning, Z-
RAYS
1. Introduction
Mainframe (mainframe machines, big iron computers) are a class of computers used
mainly by large organizations for critical applications like financial, statistical. Today,
this term (name) is usually related to computers compatible with the IBM System/360
line introduced in 1965 [1]. There is no formal definition of a mainframe. Such
machines run uninterrupted for very long periods of time and are used everywhere
when the high availability is required because any downtime would be costly or even
catastrophic. They can run in parallel different instances of operating systems thanks to
the technique of virtual machines. They are not a supercomputer [1].
In this paper we focus on a series of preliminary studies and analyses to prepare:
1
Corresponding Author, Alicja Gerka, Department of Complex System, Rzeszów University of
Technology, Al. Powstańców Warszawy 12, 35-959 Rzeszów, Poland; E-mail: a.gerka@prz.edu.pl.
Modern Management based on Big Data II and Machine Learning and Intelligent Systems III
A.J. Tallón-Ballesteros (Ed.)
© 2021 The authors and IOS Press.
This article is published online with Open Access by IOS Press and distributed under the terms
of the Creative Commons Attribution Non-Commercial License 4.0 (CC BY-NC 4.0).
doi:10.3233/FAIA210236
96