Sounds like there may be routing trouble between your ISP and the Linode datacenter (unless you're getting a _lot_ of reports from users). During the downtime, get a traceroute to your server and report it in a ticket. You might also want to try requesting a migration to another datacenter, where you might have better luck.