This was somewhat expected; almost anticipated even, just as that last comment indicates (watch the very bottom of this remark).

We have got to get the hosting sorted out because the site has outgrown itself and it cannot cope too well with Slashdot/Digg/other such effects, especially when these occur simultaneously. I tried optimising the PHP files a little bit in order to reduce the load, but to no avail. The site went down at around 5AM GMT.

The server is decent and it can cope with the usual traffic of a few hits per second, but when thousands of people arrive within a short period of time, then all those PNG files elevate throughput and the many database queries which make up detailed pages just bring the server to its knees and it eventually collapses.

Eric Meyer had an interesting discussion a couple of years ago about handling spikes by borrowing CPU cycles from the local desktops, but it seems complex and unreasonable. It's truly a shame that at the most critical times we lose valuable and new readership. It has become a pattern and there's no simple solution in sight, so if you have advice to share, we would appreciate it enormously. ⬆