I did some research on rete tree. I think what i really needs is to remove each node from tree after evaluation ( Ex: When one book is processed that and all the subsequent parts of that tree should be removed ). I tried this using "retract " modifier. But this takes more time than previous execution. According to my requirement I have to keep rule for each book separately. Therefore I can't remove those rules. Anyway you have mentioned , agenda events don't give rule execution time. So how can one measure rule execution time ? -- View this message in context: http://drools.46999.n3.nabble.com/How-to-avoid-stateless-session-rule-fir... Sent from the Drools: User forum mailing list archive at Nabble.com.

It sounds to me like one of the following: A poor requirement from someone who doesn't understand the technology. An overly literal interpretation of a requirement which was not intended to be technical. I suspect that you may well need to go back to whoever provided these requirements and work out which are hard technical requirements and which are not. And *why*. You may need to renegotiate with them with respect to what you are building. If they are any good, I would expect them to appreciate being told there are problems with their requirements now, rather than discovering in a couple of months time that the application needs to be re-written. A rules engine based application can be very fast at providing responses, but as you have found, a poor design can ensure that it seems like the slowest thing in the world. And it seems that a lot of developers go down a wrong path which requires a complete re-write of an application. With what you seem to have, there are no simple performance tweaks that will speed it up enough. It seems that it needs a fundamental rethink. One of the most important aspects of rules engine development is understanding the business intent and being able to work out how to achieve the goal effectively. As Wolfgang mentions, I reckon you would be better off providing some details of the problem so that we might be able to help you find a sensible solution. Otherwise it's looking to me like we would be helping you waste time on an inappropriate solution. Steve

Only a few minutes of our time spent responding I suspect. I was more concerned that you could easily spend days developing an inappropriate solution which would need to be re-written. :) As a rule, when asking questions here, it's far better to explain the problem you're trying to solve. Without that context, it's very difficult for anyone to provide useful answers. On a side note, when you do have a large number of 'static data' facts to work with, and want to make multiple queries, you would probably be better off using a stateful session. That way, you take a hit for compiling rules and inserting data facts up front. Subsequently you can insert a query fact, and get a response fast. If you're just validating something then you can retract it. Using this technique I have reduced application request processing times from >30 seconds to a few hundredths of a second. Steve On 7 Aug 2013, at 13:10, viraj <mayuraviraj(a)gmail.com&gt; wrote: