Work-In-Progress: Fault Tolerance in a Two-State Regularity-Based Checkpointing System
Elena Torre, Albert M. K. Cheng, Guangli Dai, Pavan Kumar Paluri · 2021
Embedded real-time virtualized systems serve a wide range of functions for many important industries. They can encompass multiple independent applications sharing limited computational resources. Many models have been introduced to ensure reliability and energy efficiency for these systems. Hierarchical Real-Time Scheduling (HiRTS) is a framework to enable the sharing of resources. It is used alongside the Regularity-Based Resource Partition model (RRP) to achieve transparent scheduling. A checkpointing system is an effective method to resolve transient faults. However, checkpoint insertions are known to incur high time and energy overheads. This paper proposes a two-state regularity-based checkpointing model within the HiRTS framework. It will ensure fault tolerance when scheduling independent, mixedcriticality real-time task sets on limited resources. By reducing checkpoint insertions before the first fault, the system will achieve higher utilization and less overhead while still ensuring fault tolerance. The simulation-based experiments presented suggest the model could offer a significant increase in utilization and reliability. They will be used as a guideline for future simulations using more complex schedules.