Can hadoop be used as a back-end for satisfying request originating from the browser?

[UsefullyWastedIce <pa...@gmail.com>]

The system we are trying to build will have its front-end accessible though a
web interface.
When a request is made, a few gigabytes of XML data will need to be pulled
from a server, and then saved in some intermediate state before being
returned to the user. Only a snapshot will be sent all the way up to the
browser, maybe the first 500 rows along with a total amount of rows. 
The user can then proceed to work with that data by filtering it based on
some criteria, as well as possibly aggregating it with some other data. 

>From everything I've read about hadoop, it doesn't really seem that it will
work well for us, because it's really meant more for batch processing, with
jobs taking minutes, or even hours to complete. Not seconds like we are
trying to achieve. 

Is there anyone out there who's used hadoop to process data on the fly, in
almost "real time"? I haven't managed to find anything like it. 
-- 
View this message in context: http://old.nabble.com/Can-hadoop-be-used-as-a-back-end-for-satisfying-request-originating-from-the-browser--tp27243987p27243987.html
Sent from the Hadoop core-user mailing list archive at Nabble.com.