Here we build on the end of lecture 5 and start to think about new kinds of platform services that can make a system more autonomic.

One, inspired by lecture 5, would be a system-wide fault detection service.  We’ll build one later in the course.  Then we talk about other autonomic services.

Gossip is a powerful tool in building such services.  We introduce gossip and show how it can be used to send event notifications (such as “node A has crashed”) very robustly.