Database Theory

I was wondering if anyone would care to explain some database theory to me. I've made a DLL for working with a database I made using binary files and stuff, and I'm wondering how other people do it so that I can make it better:).

If its necessary, the current info is this: currently, its only for one table really, because the thing I was making it for only needed one table. The first 4 bytes are a long specifying the number of records. The next four bytes are a long specifying the number of fields per record. The next four bytes after that are a pointer to the location in the file where the index starts. The index consists of 14 bytes per record, and contains a 4-byte pointer to the location in the file where the record is, and a 10-byte binary string which is used as an internal, permanent index. This is randomly generated based on certain parts of the data in the record, but once generated is permanent even if the record changes. Records' numeric indices are sorted based on this ID code. The main record consists of a 4 byte long specifying the length of the total record when read, then ten-byte id code, and a series of 4-byte pointers equal to the number of fields. These pointers, when followed, lead to 4 bytes specifying the length of the data in the field, and that many characters of data. That's about it.

Just writing this has actually made me think of ways to reduce overhead, so don't reply about that;). But I'm interested in how large database systems are designed and stuff.

BTW, when I said the records were sorted by the id code, that's just the ordering of the 14-byte indexes in the index section. That's done so references can be made to the permanent id codes quickly by looking up the numeric index using a binary search.

THe Largest I've worked with are 358 GB AS/400 Database and a 66.8 GB MSSQL Server database so far. I'm sure ther are many much much much larger because these are on relativley new companies servers. Good thing disk is cheap these days!

BUMP