Nope, they deal with Algorithms in general without regard to storage. Usually when an algorithm has been developed with some sort of memory in mind it is titled as such. As an example cache oblivious algorithms are algorithms which are cache aware but oblivious to cache size (aka supposed to be good no matter the cache size).
Sqlite is the only database I can think of that is small enough to be read. If all you are interested in is algorithms applied to disk storage I don't think I would recommend writing your own database since that is going to involve writing a large chunk of code for SQL parsing, query planing, etc. Why not just write some data processing tools for data mining.
Here is an example, you can usually increase DB performance by about 15% by not allocating the first 8 MB of your hard drive. (Hint: the partition table causes misaligned reads/writes on raid systems if your partition starts at sector 1)
If you want to see the tricks of the trade, go read the TPC results, find the auditor's report and look through the source, the hardware configuration, and the software configuration.
On most systems RAM is the 5th level of storage (registers,l1 cache,l2 cache,l3 cache, ram). Just because L1-3 cache isn't addressable doesn't mean it won't have a serious effect on the performance of your code.
_An Introduction to Database Systems_ (8th Ed.) By C. J. Date http://www.amazon.com/Introduction-Database-Systems-8th/dp/0...