Intelligent Data Platform (IDP)

Using advanced analytics, machine learning and AI, the IDP can intelligently alert users to problems before they happen through alerts, graphs, maps, charts and data visualizations that enable users to quickly interpret vast amounts of stream data and ensure broadcast-quality results.

The IDP analyzes the data and creates benchmarks using descriptive analysis of broadcast metrics such as anomaly detection, encoder quality, round trip time and dropped packets to create predictive models that alert to the likelihood of certain events such as stream failure and content quality issues.

Open

The following IDP features have been implemented in ZEN Master.

Multi-Object Correlation Analysis (MOCA)

IDP’s multi-object correlation analysis automatically groups errors into incidents, by determining correlations between individual-object incidents. The incidents are reported when there are 3 or more Sources, Targets, or Broadcasters having errors at the same time that share a common resource (see shared resources definition below), and when at least 50% of the objects that share that resource are experiencing errors at the same time frame of 10 minutes. Errors that happen on these objects within 10 minutes of a previous error are grouped together into the incident. It is possible to have multiple related incidents if there is a period of more than 10 minutes without any errors. In some cases, if there are multiple shared resources, the system may also report multiple related incident records.

The shared resource can be one of the following:

• Client Network – the network where Source or Target objects are connected. Networks are identified by network ASN (Autonomous System Number), country code, and region code. For example, US_OH_123 is a network in Ohio in the United States with ASN 123.

• Server Network – the network where a Broadcaster is connected.

• Network Path – the connection between a Source and a Broadcaster.

• Broadcaster – the shared Zixi Broadcaster.  

Based on this analysis IDP provides an automatic RCA attribution, which specifies the likely root cause of the incident.

IDP displays its analysis in the following modules:

Health Score

Zixi Health Score provides real-time predictions of network health, anticipating potential offline events and unrecoverable packet loss. It also enables monitoring of historical health status to help with forensic analysis. The health score is determined based on Zixi IDP’s ML algorithms that assess over 100 measurements of the Source’s performance. ZEN Master also provides a breakdown of the factors that contributed to a particular fluctuation in the Health Score.

Insights

Insights are a set of graphs in ZEN Master that enable you to identify outliers and likely areas of instability in your system. Insights graphs are shown for Broadcasters and for Channels. These graphs make it easy to compare different workflows, showing you where you need to focus your attention. Graphs show metrics related to packet recovery as well as other Broadcaster KPIs such as CPU and memory usage.

Incidents

The Incident view groups together related errors into incidents to simplify the process of identifying the common root cause. ZEN Master identifies the object on which the initial error occurred and the network on which the channel is transmitted. It also suggests a likely cause of the incident. Incidents can be created manually and/or automatically generated based on the MOCA analysis, as described above.