Drools Performance and Maintanability

[dollanitri]

Hello,

Recently i started playing out with Drools Guvnor GUI and created couple of rules based on DecisionTable
GUI Editor and also ruleflows in eclipse.

Then integrated Drools with Java Custom Web Application and tried out executing rules on front end by changing rules on the drools guvnor at run time. All seems to be working fine as per expected.

But i have couple questions would like to ask related to performance and maintainability of the drools.

Questions:

1) does drools guvnor provides any option to make revisions for the decision tables ?
(Eg. If more than single developer working on the same decisiontable and making modifications to it then
it should allow to merge all the changes into main one.....)

2) how to measure performance of the drools to make it work and available for the large rule datasets?

3) There are options to store data in a different formats like repository (default one), Oracle, mysql.....etc
which one is better if i want to make my data backup very easily and import it back whenever i need ?

4) is it drools is good one to replace with Yasu?

[manstis]

In response to a couple if your questions:-

1) Guvnor uses an optimistic lock at the asset (e.g. decision table) level. We do not provide for merging as you describe.

3) Guvnor uses Apache Jackrabbit for persistence. Jackrabbit can be configured to use a database however the schema Jackrabbit uses is not specialised for Drools and hence querying as you might like is next to impossible.

One of the core Drools engine developers, or community, may be able to advise re: #3.

With kind regards,

Mike

sent on the move

On 23 Apr 2012 07:34, "dollanitri" <[hidden email]> wrote:

Hello,

Recently i started playing out with Drools Guvnor GUI and created couple of
rules based on DecisionTable
GUI Editor and also ruleflows in eclipse.

Then integrated Drools with Java Custom Web Application and tried out
executing rules on front end by changing rules on the drools guvnor at run
time. All seems to be working fine as per expected.

But i have couple questions would like to ask related to performance and
maintainability of the drools.

Questions:

1) does drools guvnor provides any option to make revisions for the decision
tables ?
(Eg. If more than single developer working on the same decisiontable and
making modifications to it then
it should allow to merge all the changes into main one.....)

2) how to measure performance of the drools to make it work and available
for the large rule datasets?

3) There are options to store data in a different formats like repository
(default one), Oracle, mysql.....etc
which one is better if i want to make my data backup very easily and import
it back whenever i need ?

4) is it drools is good one to replace with Yasu?


--
View this message in context: http://drools.46999.n3.nabble.com/Drools-Performance-and-Maintanability-tp3931655p3931655.html
Sent from the Drools: User forum mailing list archive at Nabble.com.
_______________________________________________
rules-users mailing list
[hidden email]
https://lists.jboss.org/mailman/listinfo/rules-users

_______________________________________________
rules-users mailing list
[hidden email]
https://lists.jboss.org/mailman/listinfo/rules-users

[Nicolas Héron]

Hello,
about your point #1, you can use status of rules and therefor only build in production if a rule is in special case, but is is tricky and there is no merge facility and possibility in Guvnor. The only way is to do it from whithin eclipse with the the guvnor plugin over webdav.
about your point #2 : I just finished a project with over 100k rules (huge decision tables). Building the projects take a few minutes (I was in a virtual environment which is not a good idea for this case), building the knowldgebase from Guvnor => 30s and getting a new knowledge session, like usual, not measurable
about your point #3 : it works fine with mysql and postgresql, as I tested it.
regards
Nicolas Héron

2012/4/23 Michael Anstis <[hidden email]>

In response to a couple if your questions:-

1) Guvnor uses an optimistic lock at the asset (e.g. decision table) level. We do not provide for merging as you describe.

3) Guvnor uses Apache Jackrabbit for persistence. Jackrabbit can be configured to use a database however the schema Jackrabbit uses is not specialised for Drools and hence querying as you might like is next to impossible.

One of the core Drools engine developers, or community, may be able to advise re: #3.

With kind regards,

Mike

sent on the move

On 23 Apr 2012 07:34, "dollanitri" <[hidden email]> wrote:

Hello,

Recently i started playing out with Drools Guvnor GUI and created couple of
rules based on DecisionTable
GUI Editor and also ruleflows in eclipse.

Then integrated Drools with Java Custom Web Application and tried out
executing rules on front end by changing rules on the drools guvnor at run
time. All seems to be working fine as per expected.

But i have couple questions would like to ask related to performance and
maintainability of the drools.

Questions:

1) does drools guvnor provides any option to make revisions for the decision
tables ?
(Eg. If more than single developer working on the same decisiontable and
making modifications to it then
it should allow to merge all the changes into main one.....)

2) how to measure performance of the drools to make it work and available
for the large rule datasets?

3) There are options to store data in a different formats like repository
(default one), Oracle, mysql.....etc
which one is better if i want to make my data backup very easily and import
it back whenever i need ?

4) is it drools is good one to replace with Yasu?


--
View this message in context: http://drools.46999.n3.nabble.com/Drools-Performance-and-Maintanability-tp3931655p3931655.html
Sent from the Drools: User forum mailing list archive at Nabble.com.
_______________________________________________
rules-users mailing list
[hidden email]
https://lists.jboss.org/mailman/listinfo/rules-users

_______________________________________________
rules-users mailing list
[hidden email]
https://lists.jboss.org/mailman/listinfo/rules-users

_______________________________________________
rules-users mailing list
[hidden email]
https://lists.jboss.org/mailman/listinfo/rules-users

[dollanitri]

Thanks for your inputs......

#1. it means if multiple users update same record (in this case decision table ) then other one changes will be overwritten, as it shouldn't be i believe.

#3. is it way to import this repository into eclipse to see data by using some plugin ?

[manstis]

#1 - Two users, (1) and (2),  open the same asset. User (1) saves OK. User (2) will be prevented from saving as their view of the underlying asset is out of date.

This is how the optimistic lock pattern works. It is ideal for low con-currency usage patterns, which is how we envisage Guvnor (assets) being used.

Whether this suits your use-case is a different question entirely.

#2 - You can use the Drools Eclipse Plugin to attach to a Guvnor repository and view it's content. This is a little out dated though and some assets (in particular the web based decision table) cannot I suspect be viewed.

On 23 April 2012 10:50, dollanitri <[hidden email]> wrote:

Thanks for your inputs......

#1. it means if multiple users update same record (in this case decision
table ) then other one changes will be overwritten, as it shouldn't be i
believe.

#3. is it way to import this repository into eclipse to see data by using
some plugin ?


--
View this message in context: http://drools.46999.n3.nabble.com/Drools-Performance-and-Maintanability-tp3931655p3931971.html

Sent from the Drools: User forum mailing list archive at Nabble.com.
_______________________________________________
rules-users mailing list
[hidden email]
https://lists.jboss.org/mailman/listinfo/rules-users

_______________________________________________
rules-users mailing list
[hidden email]
https://lists.jboss.org/mailman/listinfo/rules-users

[dollanitri]

Hi Nicolas,

Thanks for your valuable inputs.

#2, did you observered how much memory was consumed while building such huge rules ?
do we need to consider memory consumption in this case?