I'm trying to shove the University into Zope. I've got (~60K?) users handled and now I want to attack classes. We have a text file with roster data and an Oracle database of course info. My plan is to make a folder for every section of every course, and make objects or methods for every piece of data we have about each course (like "description" and "prerequisites") in these folders. This means I'll have paths like /courses/2001.1/AT/249/03/02 and that would have /courses/2001.1/AT/249/03/02/instructor even though the instructor might be acquired from a parent folder. So...my first step (late last night/early this morning) was to try to make folders for every section. We have 10,699 sections this semester, and I ran into some limits (like FastCGI timeout, then Netscape timeout, ...) in populating the ZODB. Before I just start kludging my way around this, I want to make sure there's not a better way. I'll be doing this every semester, so I want to do it well. Today I learned about XML-RPC. We're hard at work trying to get the HTTPS capability welded on to it. (We only use HTTPS for Zope.) It seems that it might be better to use this to populate the ZODB. Although I'd rather keep everything in Zope, I'm thinking I'd write a Python program that parses the text file and communicates with Oracle to get all of the data. As it's iterating over that data, it would shoot off XML-RPC transactions with Zope. There would be something on the order of 17,000 transactions. If I decide to handle rosters through the ZODB, there would be another 230,000 transactions. Is this a reasonable way to handle this? Am I going to kill the ZODB? Will I swamp it with transaction logs? I recall seeing that the ZODB isn't so great for writing. I'll be doing a lot of writing. We've used Oracle for everything, but I want so much to move completely to ZODB so that I can use ZEO. I'm willing to put in some extra effort to move down this path. Thank you. --kyler