Product Details
Analyzátor metrik monitorovaných v dohledovém centru
Created: 2022
Hruška Martin, Ing., Ph.D. (Automata@FIT)
Smrčka Aleš, Ing., Ph.D. (DITS)
Švéda Martin, Ing.
Hradský Tomáš, Ing.
metrics; logs; monitoring; data analysis; Elasticsearch; Kibana; Industry 4.0
In the result, emphasis was placed on the so-called offline analysis of metrics
and online analysis of metrics (e.g. in the form of detection of suspicious
values, in real time). Offline data analysis is implemented as a perfaide tool,
which provides a command-line interface for, e.g.: (1) trend analysis, (2)
automatic creation of configurations, (3) automatic creation of visualizations,
or (4) correlation analysis among the monitored metrics. Online data analysis is
implemented as a plugin of the Elasticsearch platform called perfcheck. The
plugin form of the Elasticsearch was chosen for its ability to process individual
records of the current state of metrics before saving them to the internal
Elasticsearch database, and at the same time for the possibility of its close
integration within the V001 result. The plugin includes a set of deviation
detection methods as well as heuristics for filtering out the part of the report
that is not satisfactorily informative for the user.
Zdrojové kódy a dokumentace se nachází v GitLab repozitáři https://pajda.fit.vutbr.cz/tacr-unis/perfaide. Do repozitáře je umožněn přístup pouze na základě smlouvy o komerčním využití výsledků projektu.