● One of the primary purposes of an HPC implementation is scalability. However, with more nodes, the probability of failure is higher. Individual components (such as processors, disks, memories, and networks) have a high mean time before failure, but a system with numerous components may fail frequently. In a system with 10000 nodes, each with a mean time to failure 106 hours, the system will have only a mean time to failure 100 hours. ● Therefore, dealing with faults is crucial for a scalable HPC application. ● One of the leading research challenges in this area is developing fault tolerance techniques that do not cause significant overheads in normal execution conditions and do not hinder the basic HPC optimizations applied. ● We propose a Dependability Library for Iterative Applications (DeLIA) in C++ for HPC environments. ○ An fault tolerance library to HPC applications which have global synchronization between all processes, obtaining a globally consistent state after each iteration (see Figure 1). ○ The focus is to deal with failures which stops all execution and preemptive environments.