I've created a Web/ASP.NET/MySQL database front end & back end that gets created from the XML logs from a previous Java based XML client-server that is not serving the company very well (the XML database bloats too quickly and the server begins to time out, which halts the production floor).
Any future data will come through the web front end or API calls to the back end, and we'll eventually (after testing) shut down the old client-server database.
We're about to go live for testing with the new database, and I'm worried about data integrity with the imported data. We ended up with a bunch of duplicate data (about 18%, see SO question about this here) which I've cleared up (or at least come up with a repeatable method to do so). I've also tested that each of the row ID's that are primary keys actually have data associated with their foreign key children (and their's have the same PK-FK row data association down to the lowest child table, hopefully that made sense). However, I've only tested that the rows contain data, not that they contain the right data.
My thought is to choose random log files from the old XML database, and make sure that its data is where it should be in the new MySQL version. Ideally I'd be able to automate this, but it's complex to break down the old logs into their constituent parts to insert into the database, but maybe not so much to simply search for the data contained in them.
I'm relatively stupid when it comes to DBA stuff, and I realize what I'm asking is kind of vague, but what I'm looking for is what would a DBA want to know to make sure they were working with reliable data? If this can be figured out and tested for (my forte), I'm going to feel a lot better about the data. The entire company will be relying on this data once it goes live, I need to get this right.