Help with peculiar DB setup

[bob bob]

Hi,
I am doing research on a large MyISAM DB. The table has about 65+ million lines and is about 14GB in size. Each record is a log entry from a network device, with typical date, time, IP_address, user, etc type details. If you imagined standard apache logs then you would not be far off.

I am the only user of the server and I tend to do complex one off queries.

My question is how to maximise performance of hardware/architecture before I start on the DB itself. Should I split the large table into smaller tables given I will often want to query all time values? Not sure on the criteria of how to split.

MySQL does not thread a single query and so I don't think number of cores or processors will help much. Is that correct? I have several Xeon servers available but can not see how they can help?

Most queries seem to be bound on read access. Is there anyway to move the how thing to memory somehow? Or will much faster disks help, maybe even Solid State?

Can someone point me to design or discussions for designing better performance for this single query scenario? Thanks very much