Architecture
This topic provides a brief overview of the Predictive Insights component architecture.
The following figure shows an overview of the Predictive Insights component architecture.
Data Source
A data source is a repository of data that is generated by an external application such as an organization's monitoring and performance management system. Predictive Insights can consume data from multiple data sources, which can be either relational database management systems or file based systems.Mediation Tool
The Mediation Tool is a graphical configuration tool that is used to define the data within each data source that Predictive Insights consumes and analyzes.Database component
The database component uses a Db2 database. The database is a repository for configuration information, data that is extracted from the data sources and anomaly information.Analytics component
The Analytics component extracts data from each data source. Initially, the Analytics component uses the extracted data to build an analytics model with data that represents the normal behavior of the system. After the analytics model is built, it compares subsequent data that is extracted with data in the model so it can identify any changes in system behavior.A single Analytics component on a dedicated server can analyze up to 350,000 Key Performance Indicators (KPIs) at a 5-minute interval and 500,000 KPIs at a 15-minute interval. You can add further Analytics servers to scale the system.
User Interface
The User Interface displays detailed information for anomalies, including a visual representation of metric behavior on a time series chart.OMNIbus ObjectServer
The Omnibus ObjectServer is a prerequisite for using Predictive Insights. The ObjectServer is a database server that stores event information that is forwarded to it by the analytics component.OMNIbus Web GUI
The OMNIBus Web GUI is a prerequisite for using Predictive Insights. The Active Event List (AEL) in OMNIbus WebGUI is an interactive Java applet for displaying alert data from the OMNIbus ObjectServer. As an alternative, the OMNIbus WebGUI Event Viewer can display alert data and does not require a Java plugin.Visualization Application
A visualization application such as Dashboard Application Services Hub is required to view events in the AEL. Using the Dashboard Application Services Hub console, an operator can launch the Predictive Insights User Interface to view and analyze anomaly data.REST Mediation Service
Use this service to send JSON data to Predictive Insights. The data can come from a direct data POST operation, or indirectly from Kafka.