I've been doing something like this for years and my databases have about 1,000,000 items that fill 20gig of mysql, One slight difference is that I check for new models every 2 minutes and when one is discovered it goes into a new product rss feed that I use as content for peripheral sites. Its not all scraping, about 3/4 if the 1.5gig of data it processes per day is xml->php->mysql without any scraping.