The data is used with 3rd party tools that scan tables, find the best seats, display statistics etc. The most popular tool is holdem manager (http://www.holdemmanager.com/). After buying data off our site you can import hundreds of millions of hands and have detailed info on almost every player on a poker site (how often they raise, fold, call etc etc). If you also scan tables and sit with the "fish" (recreational players) then you will win a lot more. One day I'll write a blog post about what online poker is really like (game theory, tools, and statistics).
On the tech side its some fully reversed clients that we just have linux command line clients which connect to sites, and other sites its a ton of Windows XP VM's which open tables and observe. At the heart of it all we have some command servers which handle distributing the table load, parsing and aggregating all the data etc. Each day we "mine" over 10 gbs of data zipped and at peak times can be watching over 10,000 poker tables. It's pretty nuts that it somehow all works.
Yeah, a few sites have made it so you can't observe tables but most sites allow it. You'd have to ask them why but I assume that casual players like to observe before depositing and they don't want to scare away the casual players (who are the life blood for a poker site).
On the tech side its some fully reversed clients that we just have linux command line clients which connect to sites, and other sites its a ton of Windows XP VM's which open tables and observe. At the heart of it all we have some command servers which handle distributing the table load, parsing and aggregating all the data etc. Each day we "mine" over 10 gbs of data zipped and at peak times can be watching over 10,000 poker tables. It's pretty nuts that it somehow all works.
And yeah, lots of typos :-O