Remote storage with byzantine servers

We consider the problem of providing byzantine-tolerant storage in distributed systems where client-server links are much thinner and slower than server-server links. We provide storage algorithms that are unique in two ways. First, our algorithms take into consideration the asymmetry in network connectivity by minimizing client-server communication. To provide this property, we rely on a small amount of partial (eventual) synchrony. Second, our algorithms provide a new property called limited effect, which is important for storage systems. To provide the latter property, we use synchronized clocks, which are increasingly common due to GPS devices and NTP, even in otherwise "asynchronous systems" like the Internet. We present two algorithms called QUAD and LINEAR, which provide a trade-off between failure resiliency and efficiency. Our algorithms implement an abortable register [3], which is an abstraction used in some real storage systems, but abortable registers are weaker than atomic registers. Thus, one might wonder if we could have implemented atomic registers instead. We answer this question in the negative: we prove that there are no implementations of atomic registers that provide the limited effect property in systems with failures, even with synchronized clocks.