TheCoreWithin

http://arstechnica.com/science/news/200 ... r-logs.ars
http://dmitriwilliams.com/research.html

I've only read the summary, but that at least sounds interesting :)

I would not want to have to handle 60TB of database dumps. No, I wouldn't.
I already hate handling them if they grow beyond 5GB

(Although I have to admit respect for DB programmers. I have witnessed a quick-hack script in PLSQL run 20+ times faster in the database than a local list search/comparison after downloading the data.)

--G