There's a significant number of companies who build data warehouses where the primary source of data is file feeds. This is batch processing on the sending and receiving end.
Even with a trickle-fed EDW, data marts traditionally are static for a period of time (daily, weekly, monthly, etc) so that analysis can be done on static data. Again, the refresh would be a batch process.
For those who haven't read Kimball, please read this before you discard him as yesterday's news wrt real-time:
http://www.rkimball.com/html/designtips ... tip31.html
So, batch processing is dead or dying? Anyone else want to jump in here or do I start looking for another line of work? Real time has its time and place (sorry for the pun). Are you going to do real time to your ODS, EDW, and marts? Kimball comments better on it than me. I'll let everyone else out there do their research.
Kim, respectfully, I do a lot of reading. I meet onsite enough "data warehouse architects" who continually repeat the problems for which Kimball became famous for describing solutions. These "architects" don't even understand slowly changing dimensions, or how to model warehouses. They've never built aggregates, or handled volumes in the billions of rows. I'm not sick of Kimball, I wish more people understood what his books were trying to say. I've written a ton of posts, white papers, attended conferences, and given speeches and demonstrations. A tool is not going to replace due diligence on architecture, requirements gathering, and programming techniques. What I like about Kimball is his emphasis on structured, traditional approaches. Call me an old fart if you want.
If I'm sick of something, it's the lack of understanding that the ETL tool is not the silver bullet. RTI and PX doesn't change anything. Technology is an enabler, it's not the solution. A framework should be the right way to do things, then figure out how to make the technology work within the framework. You don't pick a tool or technology, then decide the right things to do.