In many businesses the monitoring system is a customer. The website is down, the phones are not ringing or orders have stopped, and the first person to notice is someone trying to buy something. Proper monitoring is inexpensive, and it changes the way IT is run: from reacting to outages to preventing them.
What to watch
- From the outside: your website, your email and your internet connections, checked every few minutes from somewhere other than your office.
- From the inside: servers, network equipment and circuits, including disk space, memory and failed backups.
- Computers: missing updates, disabled protection and failing drives, reported by a management tool on each machine.
- The room itself: temperature, humidity, water and power in the place your equipment lives. An overheated closet has ended many a working day.
- Scheduled jobs: the nightly import or monthly report that fails silently unless something checks that it ran.
Alerts people will act on
Monitoring fails when it cries wolf. If an inbox receives two hundred alerts a day, the one that matters will be missed. Set the thresholds so that an alert means something is wrong. Send each one to a named person. Decide in advance which alerts justify a phone call at night.
A sensible start
List the five things that would hurt most if they stopped. Monitor those first, from outside and inside. Add a sensor to the equipment room. Review the alerts after a month and remove the noise. Then widen the list.
Related service: IT & CTO/CIO management.