Chapter

Organic Computing — A Paradigm Shift for Complex Systems

Volume 1 of the series Autonomic Systems pp 163-177

Monitoring and Self-awareness for Heterogeneous, Adaptive Computing Systems

  • David KramerAffiliated withChair for Computer Architecture and Parallel Processing, Institute of Computer Science and Engineering, Karlsruhe Institute of Technology (KIT) Email author 
  • , Rainer BuchtyAffiliated withChair for Computer Architecture and Parallel Processing, Institute of Computer Science and Engineering, Karlsruhe Institute of Technology (KIT)
  • , Wolfgang KarlAffiliated withChair for Computer Architecture and Parallel Processing, Institute of Computer Science and Engineering, Karlsruhe Institute of Technology (KIT)

* Final gross prices may vary according to local VAT.

Get Access

Abstract

A comprehensive monitoring infrastructure is vital for upcoming heterogeneous, adaptive many-core systems. In order to enable required self-organising capabilities, a monitoring infrastructure has to provide self-awareness. Unfortunately, traditional approaches to monitoring, like hardware performance counters, lack required flexibility and are not suitable for self-organising systems.

We therefore present a flexible, hierarchical monitoring infrastructure for heterogeneous adaptive computing systems being able to provide a detailed and pristine view of the system state. On lower level, an associative counter array performs sustained monitoring of individual components of the system and provides this information to high level instances. These instances analyse and evaluate this information, and finally realise self-awareness. For this purpose, we employ a flexible, rule-based approach for runtime evaluation and classification of the system state. Further system instances, such as the task scheduler, may use the classified state as well as gathered information to realise self-x features, such as self-optimisation.

Keywords

Self-awareness Monitoring Adaptive computing