4:02 a.m.
I have been evaluating the rule engine (ie Drools 4.0.4). I have created an
application which has about 20 rules . I insert about 500 arraylists into
the working memory and each arraylist contains about 4 objects. The rule
engine picks the object that is the best match from the arraylist and
stores it .
This processing takes about 0.953 secs .
Now this very same scenario I have implemented in using java code (if
..else)...it takes about 0.016secs .
I was expecting the rule engine to give a better performance. Considering
the use of alpha node caching and other factors of the rete algorithm.
This is how one of the rules look in the rule file :
rule "P1 110"
salience 1
activation-group "1"
when
PriorityType( pricetype == "10,20,30,40,110" )
list :ArrayList( )
t:TlbObjectAttributeData( lngAttrTypeId == 110 ) from list
then
result.add(t);
end
Am I doing something wrong or this application is not harnessing the pluses
of the rule engine .
If so can you please tell me which kind of application should I build to
highlight the performance of a rule engine .
Appreciate it
Regards
Harsh Jetly
SOA
Center of Excellence
______________________________________________________________________
4:19 a.m.
Harsh Jetly wrote:
I have been evaluating the rule engine (ie Drools 4.0.4). I have
created an
application which has about 20 rules . I insert about 500 arraylists into
the working memory and each arraylist contains about 4 objects. The rule
engine picks the object that is the best match from the arraylist and
stores it .
This processing takes about 0.953 secs .
Now this very same scenario I have implemented in using java code (if
..else)...it takes about 0.016secs .
I was expecting the rule engine to give a better performance. Considering
the use of alpha node caching and other factors of the rete algorithm.
This is how one of the rules look in the rule file :
rule "P1 110"
salience 1
activation-group "1"
when
PriorityType( pricetype == "10,20,30,40,110" )
list :ArrayList( )
t:TlbObjectAttributeData( lngAttrTypeId == 110 ) from list
then
result.add(t);
end
you don't get any indexing for "from", it's just a pure
iteration and
filtration system - as it's a serious of nodes emulating this, it will
be slower than java bytecode.
Mark
Am I doing something wrong or this application is not harnessing the pluses
of the rule engine .
If so can you please tell me which kind of application should I build to
highlight the performance of a rule engine .
Appreciate it
Regards
Harsh Jetly
SOA
Center of Excellence
______________________________________________________________________
_______________________________________________
rules-users mailing list
rules-users(a)lists.jboss.org
https://lists.jboss.org/mailman/listinfo/rules-users
4:42 a.m.
Make sure when you do performance testing that you run through enough
iterations so that the engine code benefits from HotSpot.
-----Original Message-----
From: rules-users-bounces(a)lists.jboss.org
[mailto:rules-users-bounces@lists.jboss.org] On Behalf Of Harsh Jetly
Sent: 13 August 2008 11:03
To: rules-users(a)lists.jboss.org
Subject: [rules-users] Java code Vs Drools
I have been evaluating the rule engine (ie Drools 4.0.4). I have created an
application which has about 20 rules . I insert about 500 arraylists into
the working memory and each arraylist contains about 4 objects. The rule
engine picks the object that is the best match from the arraylist and stores
it .
This processing takes about 0.953 secs .
Now this very same scenario I have implemented in using java code (if
..else)...it takes about 0.016secs .
I was expecting the rule engine to give a better performance. Considering
the use of alpha node caching and other factors of the rete algorithm.
This is how one of the rules look in the rule file :
rule "P1 110"
salience 1
activation-group "1"
when
PriorityType( pricetype == "10,20,30,40,110" )
list :ArrayList( )
t:TlbObjectAttributeData( lngAttrTypeId == 110 ) from list
then
result.add(t);
end
Am I doing something wrong or this application is not harnessing the pluses
of the rule engine .
If so can you please tell me which kind of application should I build to
highlight the performance of a rule engine .
Appreciate it
Regards
Harsh Jetly
SOA
Center of Excellence
______________________________________________________________________
_______________________________________________
rules-users mailing list
rules-users(a)lists.jboss.org
https://lists.jboss.org/mailman/listinfo/rules-users
5:18 a.m.
Corneil du Plessis wrote:
Make sure when you do performance testing that you run through
enough
iterations so that the engine code benefits from HotSpot.
Something to be aware of is that there are still many optimsation we can
do to Drools. One of them is to take our interpretted Rete network and
compile it down to bytecode, so that it would resemble something closer
to hand written java. For tight looped stuff, that should give a nice
perf increase. However mostly drools is "fast enough" so we are focusing
on building out the platform and features, but we'll continue to improve
performance slowly overtime - there is still so much more we can do.
Mark
-----Original Message-----
From: rules-users-bounces(a)lists.jboss.org
[mailto:rules-users-bounces@lists.jboss.org] On Behalf Of Harsh Jetly
Sent: 13 August 2008 11:03
To: rules-users(a)lists.jboss.org
Subject: [rules-users] Java code Vs Drools
I have been evaluating the rule engine (ie Drools 4.0.4). I have created an
application which has about 20 rules . I insert about 500 arraylists into
the working memory and each arraylist contains about 4 objects. The rule
engine picks the object that is the best match from the arraylist and stores
it .
This processing takes about 0.953 secs .
Now this very same scenario I have implemented in using java code (if
..else)...it takes about 0.016secs .
I was expecting the rule engine to give a better performance. Considering
the use of alpha node caching and other factors of the rete algorithm.
This is how one of the rules look in the rule file :
rule "P1 110"
salience 1
activation-group "1"
when
PriorityType( pricetype == "10,20,30,40,110" )
list :ArrayList( )
t:TlbObjectAttributeData( lngAttrTypeId == 110 ) from list
then
result.add(t);
end
Am I doing something wrong or this application is not harnessing the pluses
of the rule engine .
If so can you please tell me which kind of application should I build to
highlight the performance of a rule engine .
Appreciate it
Regards
Harsh Jetly
SOA
Center of Excellence
______________________________________________________________________
_______________________________________________
rules-users mailing list
rules-users(a)lists.jboss.org
https://lists.jboss.org/mailman/listinfo/rules-users
_______________________________________________
rules-users mailing list
rules-users(a)lists.jboss.org
https://lists.jboss.org/mailman/listinfo/rules-users
5:17 a.m.
Why not add your ArrayList items individually into WM?
rule "P1 110"
salience 1
activation-group "1"
when
PriorityType( pricetype == "10,20,30,40,110" )
t:TlbObjectAttributeData( lngAttrTypeId == 110 )
then
result.add(t);
end
Would this not give the same result and give you the indexing Mark has
commented upon?
Cheers,
Mike
-----Original Message-----
From: rules-users-bounces(a)lists.jboss.org
[mailto:rules-users-bounces@lists.jboss.org] On Behalf Of Harsh Jetly
Sent: 13 August 2008 10:03
To: rules-users(a)lists.jboss.org
Subject: [rules-users] Java code Vs Drools
I have been evaluating the rule engine (ie Drools 4.0.4). I have created
an
application which has about 20 rules . I insert about 500 arraylists
into
the working memory and each arraylist contains about 4 objects. The rule
engine picks the object that is the best match from the arraylist and
stores it .
This processing takes about 0.953 secs .
Now this very same scenario I have implemented in using java code (if
..else)...it takes about 0.016secs .
I was expecting the rule engine to give a better performance.
Considering
the use of alpha node caching and other factors of the rete algorithm.
This is how one of the rules look in the rule file :
rule "P1 110"
salience 1
activation-group "1"
when
PriorityType( pricetype == "10,20,30,40,110" )
list :ArrayList( )
t:TlbObjectAttributeData( lngAttrTypeId == 110 ) from list
then
result.add(t);
end
Am I doing something wrong or this application is not harnessing the
pluses
of the rule engine .
If so can you please tell me which kind of application should I build to
highlight the performance of a rule engine .
Appreciate it
Regards
Harsh Jetly
SOA
Center of Excellence
______________________________________________________________________
_______________________________________________
rules-users mailing list
rules-users(a)lists.jboss.org
https://lists.jboss.org/mailman/listinfo/rules-users
5974
days inactive
5974
days old
4 comments
4 participants
participants (4)
-
Anstis, Michael (M.) -
Corneil du Plessis -
Harsh Jetly -
Mark Proctor