Specification and Runtime Checking of Derecho, A Protocol for Fast Replication for Cloud Services

Kumar Shivam, Vishnu Paladugu, Yanhong A. Liu

Reliable distributed systems require replication and consensus among distributed processes to tolerate process and communication failures. Understanding and assuring the correctness of protocols for replication and consensus have been a significant challenge. This paper describes the precise specification and runtime checking of Derecho, a more recent, sophisticated protocol for fast replication and consensus for cloud services. A precise specification must fill in missing details and resolve ambiguities in English and pseudocode algorithm descriptions while also faithfully following the descriptions. To help check the correctness of the protocol, we also performed careful manual analysis and increasingly systematic runtime checking. We obtain a complete specification that is directly executable, and we discover and fix a number of issues in the pseudocode. These results were facilitated by the already detailed pseudocode of Derecho and made possible by using DistAlgo, a language that allows distributed algorithms to be easily and clearly expressed and directly executed.

Knowledge Graph



Sign up or login to leave a comment