Uma Comparação de Interfaces de Programação para Memória Persistente

• Processo: 23/05019-3
• Modalidade de apoio: Bolsas no Brasil - Iniciação Científica
• Vigência (Início): 01 de maio de 2023
• Vigência (Término): 30 de abril de 2024
• Área do conhecimento: Ciências Exatas e da Terra - Ciência da Computação - Sistemas de Computação
• Pesquisador responsável: Alexandro José Baldassin
• Beneficiário: André Mendonça Morato Pupin
• Instituição Sede: Instituto de Geociências e Ciências Exatas (IGCE). Universidade Estadual Paulista (UNESP). Campus de Rio Claro. Rio Claro , SP, Brasil
• Vinculado ao auxílio: 18/15519-5 - Otimizações de desempenho para arquiteturas multicore, AP.JP2
• Assunto(s): Eficiência Linguagem de programação Usabilidade
• Palavra(s)-Chave do Pesquisador: desempenho | Linguagens de Programação | Memória persistente | Usabilidade | Sistemas de Memória, Transações

Resumo

Persistent Memory (PM) is a new memory technology which provides persistent data with byte addressability. Programming with persistent memory is somewhat different than that of volatile memory, since a crash while a persistent data structure is being updated may make it inconsistent. For instance, adding a new element to a traditional linked list will corrupt the data structure if the pointer from the previous node is updated to the new one but a crash occurs before the next pointer is updated. To prevent such scenarios, the updates must be performed atomically. Although several programming interfaces have been proposed in the literature, only Intel PMDK (Persistent Memory Development Kit) is actually associated with a commercial product, being the only one officially supported. For most applications, programming with PMDK requires using its low-level interface. It is somewhat hard to use and very prone to coding errors as described in the literature. In this scientific initiation project, we intend to investigate the most common proposed interfaces for programming with PM and evaluate its ease-of-use, as well as the resulting performance. Some data structures will be choosen and a qualitative study will be carried out.