That sounds pretty slow. Whether or not it is "too slow" depends on your data set size. How big? If this is 1M ratings, you're definitely doing something wrong. If it's 1 billion, maybe that's reasonable and you need more machines (or need to use a distributed approach).
Without saying what approach you are using, it's hard for anyone to give you ideas. On Fri, Dec 17, 2010 at 6:34 PM, Alessandro Binhara <[email protected]>wrote: > Hello everyone. > I'm with a recommendation system running on a mahout using tomcat. > The server has 8G of RAM with 10 processors. > > We have about 11 requests per second happening. > But we're losing a lot of answers, because tomcat is not able to respond to > all requests. > > Someone already made a tuning for this kind of application? > > thanks.. >
