OK, the datacentre has been unable to find the problem with the switch which is causing the server to stop responding to pings. We are going to move the server to  a new switch and this will cause about 20 minutes of downtime as the server will be powered down for the move.

This should hopefully stop all of the false outage reports.