Martin Kleppmann wrote about his encounter with the Redlock algorithm on the Redis website. He shared his insights as the algorithm claims to provide fault-tolerant distributed locks. Kleppmann pointed out that using locks merely for efficiency purposes does not warrant the complexity of Redlock, suggesting simpler alternatives. He discussed the importance of generating fencing tokens to ensure the safety of locks, highlighting Redlock’s failure to produce monotonically increasing tokens. Kleppmann also delved into the timing assumptions of Redlock, arguing that it is not suitable for environments where the system model is not synchronous. Overall, he concluded that Redlock may not be the best choice for scenarios where correctness is crucial.
https://martin.kleppmann.com/2016/02/08/how-to-do-distributed-locking.html