We monitor the following items:

• CPU load,CPU IOWait ('swapping');
• Disk usage;
• Bandwidth in/out;
• HTTP response time;
• TCP connections;
• Service availbility (e.g. HTTP, FTP, SMTP, etc) ;
• Processes running;
• Many other SNMP variables;
• Many other items measured through custom scripts.

We use MRTG for trend monitoring. The example below shows a real graph of the HTTP response time that we monitor for one of our Customers.

We combine this with proactive monitoring through Nagios. This way, when the response times grow beyond a preset limit, an SMS and email notification is sent: