Lightweight writeback for Byzantine storage systems

We present the first optimally resilient, bounded, wait-free implementation of a replicated register providing atomic semantics in a system in which readers can be Byzantine, up to servers ( ) are subject to Byzantine failures and servers do not communicate with each other. Un- like previous solutions, the sizes of messages sent to writers depend only on the actual number of ac- tive readers and not on the total number of readers in the system. Timestamps generated by our solu- tion are non-skipping and messages sent to readers and writers contain only a finite number of values, operation identifiers, and timestamps. We introduce lightweight write back, a new mechanism which en- ables readers to write back only the (non skipping) timestamp of the value read and not the value itself. This is particularly important since Byzantine read- ers that write back a value could force the servers to process infinitely large messages, whereas a non- skipping timestamp is practically finite in size. With a novel use of secret sharing techniques combined with writeback throttling, we manage to tolerate Byzantine readers without the use of any unproven cryptographic assumptions.