I'm unemployed and I can only afford a basic server on a budget of about $20 a month. I am therefore forced to make the absolute most out of very limited hardware, at least until one of my sites becomes financially productive when I will be able to scale my hosting plan up.

I don't want to crush your hopes, but even if you had the best product on the whole internet, it would still take a fair bit of luck and most likely quite some investment to get the site financially productive. Since margins today are very low for most use cases, you'll probably need a huge user base.

And to get that, you will need to have adequate hardware and bandwidth from the start - since a single link on a major new outlet or community website may send you a huge number of potential users. This is known as the Slashdot effect. If your site slows to a crawl, most of those user may never come back...

The practical upshot of being so highly restricted is that the code which runs fast on such a tiny box is going to kick arse when it gets run on a decent size box.

Maybe, maybe not. When you go to a large scale, more caching in RAM may become cheaper than more CPU power (often reversed in small scale systems).

Also, when doing large site with many write operations, referential integrity of the data may or may not become much more important. It does especially when doing something financial (starting with serving ads). So you'll need an ACID compliant database, which, when configured correctly for performance, will most of the time also be more RAM bound than CPU bound.

It wasn't until quite recently when contemplating what would be needed to run a massive-multiplayer online game

This needs much more than just a fast server. For a successful service, you'll need redundant systems all over the world, a huge amount of low-lag bandwidth, 24/7 admin support, developers who constantly fix bugs and improve content and so on. Thats why there are not that much companies who have a success on the long run, a working community and actually make money through their service.

and my path is clear to develop the game I have in mind and have wanted to get stuck into writing for a long time now.

I wrote a (relatively simple) Jump'n'Run game myself called BlinkenSisters. Something like that is a shitload of work for a single person.

I don't say it's impossible for a single person to write a working, modern, successful MMOG, but it would - in my opinion - take literally a decade or two to pull it off.

Slashdot, yeh I remember a friend of mine mentioning that, way back like 1997 I think when I was at college... scary stuff ^^

I think currently the biggest problem I'm going to face is the database rather than any code problems. I'm currently using MySQL but I've heard that it doesn't scale particularly well, hence the reason for the existence of NoSQL and the like.

Oddly enough until I wrote version 4 of aXML about 3 years ago, it didn't even have a database access system in it, it was pure flat files. I abandoned that route when I got to grips with MySQL because it seemed to be faster at the time, however I have this sneaky feeling that at some point I'm going to end up going back to flat files albeit using them in a more advanced way than I used to.

As for the MMORPG, it's not going to take me that long to build as I have various tricks up my sleeve... for instance being the only person in the world using aXML to develop with; which gives me a huge advantage in terms of development time and debugging time compared with writing lower level code.

Perl Nights there was pretty much fully functional after just a few days work, and that included stopping to have big arguments here with the monks and fixing a few subtle bugs in the new plack based re-write of my core engine and plugin set.

The other thing which will work in my favour is that the MMORPG is going to be set in space like the massively popular "ogame", and thus won't require too much in terms of graphic design work to get going.

Now that I have a server capable of serving a decent amount of traffic, and I've race tested localhost on an AMD Quad Core Phenom II with 4GB ram, I'm pretty confident I will be able to scale up faster than new users arrive unless the game proves to be phenomenally popular, in which case I'll be getting in touch with my mate at IBM to get some big iron behind it.

In the meantime my server provider supports instant scaling by adding extra nodes and node balancers so hopefully I have all the ingredients for the cake lined up now.

I think currently the biggest problem I'm going to face is the database rather than any code problems. I'm currently using MySQL but I've heard that it doesn't scale particularly well, hence the reason for the existence of NoSQL and the like.

The problem with MySQL isn't that it's slow. In my opinion it's just not a full, modern database. YMMV, though, depending on your needs. NoSQL stuff has some good and bad points. If you more or less have a big haystack of data to find your needle in, NoSQL databases are a good choice as any. If you need to have things like transaction, referential integrity and similar stuff (to have a certain confidence that your data is consistent...) you'll want to look into ACID complaint databases like PostgreSQL, Oracle, Microsoft SQL Server,...

In the case of your MMORPG, you could have the world data in some NoSQL database but put the essential user data (account, payment, billing) into PostgreSQL, for example.

Perl Nights there was pretty much fully functional after just a few days work

Don't have a link for that, so i can't talk about that. But for comparison, took me about 12 hours to develop, set up and insert the content of my Diplomatic Cables Database project, including setting up the virtual machine and installing the database.

Don't use '#ff0000':
use Acme::AutoColor; my $redcolor = RED();
All colors subject to change without notice.