In het verleden heb ik wat kleine probleempjes ondervonden met het draaien van Arch Linux op VirtualBox.
Voordat VirtualBox 2.0 uitkwam ging er iets erg fout met de (virtuele) harde schijf. Als je een beetje meer van je harde schijf vroeg dan was er een flinke kans dat dat uitliep op een kernel panic. Om het compleet te maken, logde VirtualBox al deze schijfperikelen ook nog eens naar een logbestand. Zo raakte je op de host-machine snel gigabytes aan zinloze logbestanden kwijt.
In the past I have experienced quite some annoyances with running Arch Linux on VirtualBox.
Before VirtualBox 2.0 there was something horribly wrong with the disk drivers. If you did slightly more disk I/O than average there was a fair chance you would end up with a kernel panic. Besides, VirtualBox does some spurious logging, trying to tell you that the disk does not work properly. It was very easy to lose gigabytes of host disk space because of the ever-growing logfiles.
I feel a bit ashamed about it, but after all those years of using Linux, I discovered the ionice(1) command only yesterday. Where nice(1) affects the CPU scheduler, ionice affects the disk scheduler. This article explains the command pretty well, so I'd recommend you to read it first.
In the online article Tales from Responsivenessland: Why Linux Feels Slow And How To Fix That it is suggested to lower your swappiness in order to obtain more responsive applications in Linux. The main reason for this suggestion was "Screw data, prioritize code". So application's code won't get swapped away and is quickly accessible when the user needs it.
FreeBSD 6.2 sometimes stops booting the kernel, with the last message on screen:
This was a SMP kernel. As soon as I switched back to a GENERIC kernel, I haven't seen this behavior anymore.
Something to investigate better whenever I have more time, but it's a good temporary fix 
Update 28 March 2008:
For an assignment we have to write a module to read a file from the filesystem and make a virtual filesystem for it. Good, with this sentence we continue for the people who still care, since all kernel developers have clicked away this page. Screaming.