S-TRAP: Optimization and Evaluation of Timely Recovery to Any Point-in-time (TRAP)

Chao Wang, Zhanhuai Li, Na Hu, Yanming Nie

This paper presents S-TRAP, a novel block-level CDP (Continuous Data Protection) recovery mechanism based on TRAP. In accordance with a certain time interval L , S-TRAP breaks down the parity chain of TRAP and generates new sub-chains. Besides, S-TRAP introduces previous block cache which can reduce the negative impact on the primary storage system. Both mathematical analysis and experimental evaluation demonstrate that S-TRAP not only has the advantage of high recovery efficiency and reliability, but also further reduces the parity storage usage. Even more important, S-TRAP reduces the negative impacts on primary storage system performance to a large extent.