willis@violet.berkeley.edu (Willis Johnson) (06/11/88)
I'm in the early stages of designing a db system which will eventually be run on a Prime mini under the PRIMOS operating system. I've heard that the most efficient way to store my data is in one enormous file (about 3 to 5 Meg). The system I have been asked to study as a possible model does it this way. Using one giant file strikes me as dangerous from a data integrity point of view and slow when modifying the data. Is there something about the way the OS handles file I/O that makes this more reasonable? As I'm just beginning what looks like at least a year of work, I'd appreciate any pointers or references. Willis Johnson willis@violet.BERKELEY.edu