Knowing the state.
WHAT TO MONITOR
Whether services are available Whether they perform acceptably Resource usage against limits Spend Errors
HOW TO MONITOR AVAILABILITY
External checks, from outside the service.
WHY EXTERNAL
A service cannot reliably report its own unavailability.
WHAT TO ALERT ON
Unavailability Errors rising Usage approaching limits Spend above threshold
WHERE TO SEND ALERTS
Somewhere independent of the service being monitored.
THAT POINT
An alert about a mail service failing, delivered by that mail service, does not arrive.
WHAT TO CHECK REGULARLY
Provider status pages, when something seems wrong Your own dashboards
WHAT NOT TO DO
Set alerts so sensitive they are ignored Monitor extensively and review nothing
WHAT TO REVIEW MONTHLY
Availability experienced Performance trends Cost trends
WHAT TO RECORD
Incidents and their causes.
