Light Weight Thread Level Health Check Framework for Highly Available Systems

Currently in HA (Highly Available) [2] applications health is monitored using SAF (Service Availability Forum) framework [1]. There are multiple threads spawned by each of the HA application, these threads are not being monitored for: Thread is dead, Thread is blocked on a resource [7], Thread queue is full and it's not able to receive messages. If these threads are to be monitored then they have to be registered with the SAF framework, which is a bulky proposal as the threads are expected to do minimal overhead tasks. This necessitates the need for a framework which will enable applications to monitor the threads. The solution proposed in this paper can be implemented as a generic framework (minimizing the application overhead) that is applicable to both hand coded and model (auto code generated) threads. A comparative study of the different approaches available for thread level health check monitoring has also been provided to illustrate the suitability of the proposed approach for highly available applications.