The paper discusses our practical experience and theoretical results in investigating the impact of consistency on latency in distributed fault tolerant systems built over the Internet. Trade-offs between consistency, availability and latency are examined, as well as the role of the application timeout as the main determinant of the interplay between system availability and performance. The paper presents experimental results of measuring response time for replicated service-oriented systems that provide different consistency levels: ONE, ALL and QUORUM. These results clearly show that improvements in system consistency increase system latency. A set of novel analytical models is proposed that would enable quantified response time prediction depending on the level of consistency provided by a replicated system.