Learning by failing
Avishai Ish-Shalom (@nukemberg)
"memcached latency" = Latency + Network + client GC + client queueing + ....
If wait for errors, good luck
*As much as possible
Monitoring systems need to be more available and scalable than the systems being monitored.
- Adrian Cockcroft
Monitoring system should be
Uptime is an illusion created by lack of monitoring