I have been working in Blaze 6.9 version. Do any one have any idea how many rules can be executed in Blaze without performance issues ?
What do you mean by "performance issues"? Total time to complete a request? Processing time in a rules service as opposed to some other implementation mechanism?
A rules engine like Blaze Advisor is optimized for rules. Performance will generally degrade with increasing data, not rules (though certain kinds of conditions do seem to make the RETE network explode, and thereby cause degrading performance).
You will certainly find a limit to the actual number of rules you can squeeze into a rules service, but given the range of deployment/execution options available, that should for the most part be a theoretic exercise, given a moderately reasonable service design.
Could you perhaps describe your situation in more detail? What kinds of rules/services are you creating? What is your current number of rules? What is your historical/current performance numbers? Anything in particular you worry about?
I apologise for not being very clear . Its a decision table and i am executing only that , currently i have about 1500 rules ( meaning the number of rows ) . If say i hit about 10,000 rules (rows) will it impact my processing time >
Assuming that , how many rules that can be added at maximum to a decision table ?
1500 rules is not a lot. 10 000 rules is still ok. Will 10 000 rules take longer to process than 1 500 rules? Certainly!
If 10 000 rules becomes a problem (or 15 000, 30 000, 50 000...), the question would then be: Do you really need all the rules to be in the same ruleset, or are you able to split it into smaller chunks that can be selected based on context?
If splitting is not an option:
Do you execute your rules in Sequential or Default mode? If you use sequential mode, and you run into performance trouble, you can try compiled sequential mode (separate license). If you use Default mode, you can try optimized inference.
There's been talk about being able to add hundreds of thousands of rules to a decision table. I've never tried it. Someone from FICO could perhaps join in and explain the theoretical limits to the number of rules, but I suspect the management interface will crumble before the execution engine (though I can't prove it).
Retrieving data ...