Os sistemas com memória não volátil (NVM) precisam ser consistentes à falhas. Dentre os principais desafios está criar mecanismos de checkpointing viáveis em termos de desempenho e usabilidade, para isso é necessário reduzir o número de escritas na NVM, pois o aumento excessivo gera maior uso de largura de banda e, consequentemente, degrada o desempenho. Neste trabalho é proposto DONUTS, um mecanismo transparente ao software que gera épocas dinâmicas por meio de checkpoints integrados à política de substituição de cache. Comparado ao sistema anterior de melhor desempenho, o método proposto reduziu a quantidade de escritas na NVM em 53,8%, fornecendo consistência à falhas com menos de 1% de overhead de tempo de execução.

Fault injection has been an important mechanism to test the dependability properties of a system. Through this mechanism, it is possible to analyze the behavior of a computer program in case of anomalies and to obtain useful statistics to measure the effectiveness of techniques for fault tolerance. In areas such as telecommunications, aviation and finance, the use of fault tolerance is a common practice, although, in the development of simple embedded systems it usually does not occur. Due to this reason, there is a need to create tools for noninvasive tests that allow simulating faults without dramatically increasing the system complexity. This article presents some of the techniques often used to impose fault tolerance and describes a system developed for fault injection, which operates by inserting faults in certain memory regions to change the data and to cause crashes in a rapid prototyping platform for microcontrollers. At the end of this article the results are presented.

This paper describes the implementation of fault tolerance techniques (based on data and processing redundancy) in programming of a rapid prototyping platform using microcontrollers. To evaluate performance of these techniques was used a fault injector software and a weather station system as a case study. Experiments simulated faults in sensor readings and faults in SRAM memory regions of the weather station. Finally, the fault-tolerant system performance is presented in comparison with non-fault-tolerant system, considering incidence of failures, processing time, memory and power consumption.

Non-volatile memory (NVM) is an emerging technology being explored as an alternative to DRAM main memory in computing systems because of its persistence, higher storage density, lower energy consumption, and access latency close to DRAM. However, persistent memory systems must ensure data consistency on system failures, a property known as crash consistency. One of the main challenges in these systems is creating efficient checkpointing mechanisms in terms of performance and usability. Thus, it is necessary to remove persistence from the critical execution path and reduce the excessive number of writes to NVM caused by logging operations, resulting in increased memory bandwidth usage. Another limitation is that most proposed mechanisms restrict application source code to programming interfaces based on transactional models, typed as software-based approaches. These factors make it challenging to adopt NVM systems. This article presents a software-transparent mechanism based on dynamic epochs with logging operations via processing-in-memory and checkpoints integrated into the cache replacement policy. Compared to the previous best-performing system, our strategy reduces 50.6% of writes to NVM. Furthermore, it does not increase the average memory bandwidth usage, providing crash consistency with less than 2% runtime overhead.