Unavailability of a large segment of the network. In this case, users from one network segment can’t get access to websites in the other segment (eg, due to routing errors). It’s easy enough to install special utilities, but you can’t usually perform anything in this case. And the problems of this level (global) typically affect hundreds of thousands of sites, and are being solved them quickly.
Unavailability of a hoster (small segment of the network). In this case, the problems usually lie on the side of the provider (and therefore can be corrected not so fast). Accurately set the frequency and nature of these problems manually is practically impossible, but it is relatively easy with tools for automatic monitoring.
Great timeout. In this case, in fact, the site works. And open to all users. But not all users can wait for it. It is also possible to check (server timeout with WEBO Pulsar (which can send you SMS or e-mail as soon as your website response time is too large), and the total download time with www.webpagetest.org. If you page generation time on the website is too large, you should apply any of optimization techniques and maybe order Fast website. Otherwise there will be big losses among target users.
HTTP error on website. Quite often it happens that due to heavy load (or service provider hardware failures) website starts with an error (500, 502, 503 or similar). Tracking of such things and in-time moving the website to a safer place is also possible, but this of course requires special attention. WEBO Pulsar would be the best for this purpose (to track when, what time, and with what frequency). With it all unprepossessing “kitchen” ISP come out and you’ll know exactly what is happening with your site.