Practical State Machine Replication with Confidentiality

We address the problem of how to store and process data privately in cloud environments that employ state machine replication. We show that the only known solution to the problem (Yin et al., SOSP '03) is potentially susceptible to attacks. We then present a new protocol that is secure in the stronger model we formalize. Our protocol uses only efficient symmetric cryptography, while Yin et al.'s uses costly threshold signatures. We implemented and evaluated our protocol. We show that our protocol is two to three orders of magnitude faster than Yin et al.'s, which is less secure than ours.