Today, we released version 1.3 of our Kieker framework for continuous monitoring and analysis of software systems.

New Features:

• Added support for periodic samplers
• Added support for monitoring system-level statistics based on the Sigar API
• MonitoringController no longer restricted to singleton instance
• Added support for custom time sources
• Improved configuration management
• Added named pipe reader/writer (for passing monitoring records directly within same JVM)
• MonitoringController accessible as JMX MBean

API changes

• The format of the kieker.monitoring.properties changed!
• Custom writers will need to be adjusted to a changed interface
• Minor changes in the MonitoringController interface

• Minor bug-fixes and refactorings
  (e.g., renaming Kieker.Tpmon/Tpan->Kieker.Monitoring/Analysis )
• Further enancements to trace analysis tool
  (e.g., analysis & visualization features providing software assembly- & deployment-level views)
• Improved documentation, e.g., brand-new user guide with examples!
• Improved Windows support (e.g., wrapper scripts)

Trace analysis tool improved and extended:

• New internal meta-model, scalable on-the-fly trace reconstruction, …
• Analysis features: e.g., HTML output of reconstructed system model, computation of trace equivalence classes
• Visualizations, e.g., hierarchical dependency graphs (operation-, component-, container-level), sequence diagrams, call graphs

Considerable improvements have been made to Kieker’s architecture, e.g.,

1. a generalized/extensible monitoring record model (more than executions);
2. generalized/extensible writer model (more than database and filesystem); and
3. we introduced a common data model for monitoring and analysis.

This release is the first to include the analysis component Tpan (visualization features currently restricted to sequence diagrams and dependency graphs)!