Software is everywhere and used more and more every day to conduct our business, maintain our links with society, and to enhance our lives. We want that software to process our requests whenever we want, regardless of whether the software controls web servers, ATMs, the internet or the phone system. We don't want to wait in mid-transaction until someone in some far-away place turns the power on and off to reset the software.

Software that tolerates faults has fewer user impacting failures. When software is designed to detect, isolate, contain and process faults and the errors that they cause then the software system's availability and reliability is increased. Non-functional requirements for availability, reliability and fault tolerance have been important in some application domains for years, and are continually becoming more important in others.

This session will introduce many patterns for increasing software's fault tolerance and hence its reliability and availability. These patterns have been used in many software systems, from many different application domains, for many years. The patterns are of the scope that individual architects and designers can include them to build fault tolerance into their software systems.