Collect runtime metrics for containerized applications

[dimitar-dimitrow]

To enable an AIoT applications lifecycle management, it's essential to provide information in different aspects of the context these applications are deployed to, running in and managed by - i.e. provide metrics.

In the scope of the containers management runtime these could be outlined into two levels/contexts:

• of the runtime itself
• per container

Environmental metrics data can be classified into the following categories of interest applicable for each of the levels above:

• CPU
• Memory
• IO
• Disk Access
• etc. ...

Note Metrics reporting is not part of the OCI specification. Whatever API is designed, it must be ensured that:

• it's aligned with existing such
• it's applicable in the pluggable context of the Kanto containers management engine

Acceptance:

• All required metrics are provided for each of the levels
• The mechanism (push/pull) for acquiring the metrics of interest (filtering) is defined and available via the public APIs for external applications/services

Tasks:

• Implement collecting runtime metrics for containerized applications #32
• Provide unit tests for collecting runtime metrics for containerized applications #34
• Update user docs in regard to container metrics kanto#121