Re: [rules-users] How to make this rule more memory efficient?
by Greg Barton
It can handle it. The limiting factor is available heap.
GreG
On May 4, 2010, at 10:55, Ryan Fitzgerald <ryan.fitzgerald(a)ericsson.com> wrote:
Hi Thomas,
You are correct in your interpretation of what I want to do.
I did consider adding the mappings as their own facts but I was concerned that the sheer number of them would overload drools. For each Cell instance, it can refer to up to 60 ProxyCell instances. With 20,000 Cell instances (in my proto-type), that means 1.2 million mappings.
Can drools handle these kinds of numbers of facts?
Thanks,
Ryan.
-----Original Message-----
From: rules-users-bounces(a)lists.jboss.org [mailto:rules-users-bounces@lists.jboss.org] On Behalf Of Swindells, Thomas
Sent: 28 April 2010 10:17
To: Rules Users List
Subject: Re: [rules-users] How to make this rule more memory efficient?
If I understand the following
What I want to do is find out where I have a reference from any
instance of Cell - cell1 - to any instance of ProxyCell - proxycell2 -
but am missing a reference from cell2 to proxycell1 where proxycell2
is a representation of
cell2 and proxycell1 is a representation of cell1.
What I think you saying is that if
C1 -> P2 then there must be a C2 -> P1
Where C2 == P2.cellId and P1 == C1.cellId.
I've not worked this through properly (It's still too early in the working day) but do you may make a win by inserting the mappings into working memory as their own facts (and so trading memory for processing time). You can then reason over these relationships which should reduce the number of combinations that are made.
So have a rules like the following:
Rule "createRelationship" salience [high]
$cell1 : Cell()
$proxy2 : Proxy($cell1.relationships contains this.name) then insert new Relationship($cell1.id, $proxy2.id); end
Rule "check inverse" salience [low]
$relationship : Relationship()
$cell1 : Cell(id == $relationship.cellId)
$proxy2 : Proxy(id == $relationship.proxyId) not exist Relationship(cellId == $relationship.proxyId, proxyId == relationship.cellId) then //raise warning or whatever end
Hopefully this does what you wants and should be more efficient processing wise and activation count wise.
Thomas
-----Original Message-----
From: rules-users-bounces(a)lists.jboss.org [mailto:rules-users-
bounces(a)lists.jboss.org] On Behalf Of Ryan Fitzgerald
Sent: 28 April 2010 09:46
To: Rules Users List
Subject: Re: [rules-users] How to make this rule more memory efficient?
Thanks Wolfgang.
Should I conclude therefore that high memory usage is unavoidable due
to the number of facts that must be processed? I was hoping that there
might be a clever way of writing the rules so that the rules engine
can minimise the network it generates - even if it takes a little longer to execute.
All of the facts are read from a database and thus loaded into working memory.
My fallback position is to not load all these objects into working
memory but instead embed into a rule, a service call that queries the
db for all inconsistencies (effectively using SQL instead of a drools
rule). However, since I will have other rules that operate on these
objects in working memory, I was hoping to use these facts for all rules.
/Ryan.
-----Original Message-----
From: rules-users-bounces(a)lists.jboss.org [mailto:rules-users-
bounces(a)lists.jboss.org] On Behalf Of Wolfgang Laun
Sent: 27 April 2010 14:59
To: Rules Users List
Subject: Re: [rules-users] How to make this rule more memory efficient?
If there are n Cell facts,
$cell1 : Cell()
$cell2 : Cell()
needs to create n*n pairs in the network before any reduction may set in.
Using
$cell2 : Cell(this != $cell1)
reduces this by n, so we're still quadratic. Something like
$cell1 : Cell( $id1 : id )
$cell2 : Cell( id > $id1 )
reduces it to n*(n-1)/2 which is less than 50% of the original, but
still quadratic.
-W
2010/4/26 Swindells, Thomas <TSwindells(a)nds.com>:
You should probably restrict it so that
$cell2 : Cell(this != $cell1)
Thomas
From: rules-users-bounces(a)lists.jboss.org
[mailto:rules-users-bounces@lists.jboss.org] On Behalf Of Ryan
Fitzgerald
Sent: 26 April 2010 15:34
To: rules-users(a)lists.jboss.org
Subject: [rules-users] How to make this rule more memory efficient?
Hi,
Can anyone advise me on how to make a drools rule more memory efficient?
Here is the problem:
I have a Cell and a ProxyCell object classes.
The ProxyCell represents the Cell when their internal ID's match.
Each Cell and ProxyCell however has a unique name (not same as ID).
A Cell can reference (by name) a ProxyCell (as long as the ProxyCell
does not represent that actual Cell - which would effectively be a
self-reference and is not allowed).
What I want to do is find out where I have a reference from any
instance of Cell - cell1 - to any instance of ProxyCell - proxycell2
- but am missing a reference from cell2 to proxycell1 where
proxycell2 is a representation of
cell2 and proxycell1 is a representation of cell1.
Here is the rule I have written for it:
rule "Check consistent references"
when
$cell1 : Cell()
$cell2 : Cell()
$proxycell1 : ProxyCell ( id = $cell1.id, $cell2.references
contains this.name )
$proxycell2 : ProxyCell ( id = $cell2.id, $cell1.references
not contains this.name )
then
//report an error.....
end
I have 10,000 instances of Cell and 10,000 instances of ProxyCell in
working memory. For each instance of Cell, it can have references to
60 different ProxyCell instances. Loading the Cell and ProxyCell
instances into working memory is not a problem. However, when I try
to run this rule above, the memory quickly goes above 1GB and I
eventually get
an out of memory error.
I was wondering if there is a better way to structure or write this
rule so that it doesn't use so much memory.
Thanks,
Ryan.
________________________________
********************************************************************
**
**************** This message is confidential and intended only for
the addressee. If you have received this message in error, please
immediately notify the postmaster(a)nds.com and delete it from your
system as well as any copies. The content of e-mails as well as
traffic data may be monitored by NDS for employment and security
purposes. To protect the environment please do not print this e-mail
unless necessary.
NDS Limited. Registered Office: One London Road, Staines, Middlesex,
TW18 4EX, United Kingdom. A company registered in England and Wales.
Registered no. 3080780. VAT no. GB 603 8808 40-00
********************************************************************
**
****************
________________________________
This message is confidential and intended only for the addressee. If
you have received this message in error, please immediately notify
the postmaster(a)nds.com and delete it from your system as well as any
copies. The content of e-mails as well as traffic data may be
monitored by NDS for employment and security purposes.
To protect the environment please do not print this e-mail unless necessary.
An NDS Group Limited company. www.nds.com
_______________________________________________
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
_______________________________________________
rules-users mailing list
rules-users(a)lists.jboss.org
https://lists.jboss.org/mailman/listinfo/rules-users
**************************************************************************************
This message is confidential and intended only for the addressee. If you have received this message in error, please immediately notify the postmaster(a)nds.com and delete it from your system as well as any copies. The content of e-mails as well as traffic data may be monitored by NDS for employment and security purposes. To protect the environment please do not print this e-mail unless necessary.
NDS Limited. Registered Office: One London Road, Staines, Middlesex, TW18 4EX, United Kingdom. A company registered in England and Wales. Registered no. 3080780. VAT no. GB 603 8808 40-00
**************************************************************************************
This message is confidential and intended only for the addressee. If you have received this message in error, please immediately notify the postmaster(a)nds.com and delete it from your system as well as any copies. The content of e-mails as well as traffic data may be monitored by NDS for employment and security purposes.
To protect the environment please do not print this e-mail unless necessary.
An NDS Group Limited company. www.nds.com
_______________________________________________
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
15 years, 11 months
Drools + ETL Process
by andre.fonseca＠mail.com
Hi all,
I am using Drools 5.0 with Tomcat 6 and MySQL.
Now I need to insert data in the drools database but I DON'T want to use the Guvnor interface.
This is because in the future we'll need to use a kind of ETL process to populate the drools database.
Should I create a java class with JDBC access and perform a SQL direct to that tables?
I did the reverse engineering of the drools database but I could not understood yet where the data are persisted.
Any advice should be very helpful.
Thanks in advance.
15 years, 11 months
Human Task with a graphical representation
by rudolf michael
Hello there,
I was able to draw an "Incident on Workstation Process" using Drools flow
and i am going to test it today.
I have a Timer on every Human task where i need to log an OLA Exception when
the timer is due. this is easy but what is missing a graphical
representation for this.
is there a way to show/draw this information in my eclipse flow plung in so
for the business analyst to see it?
best regards,
Rudolf Michael
15 years, 11 months
How to make this rule more memory efficient?
by Ryan Fitzgerald
Hi,
Can anyone advise me on how to make a drools rule more memory efficient? Here is the problem:
I have a Cell and a ProxyCell object classes.
The ProxyCell represents the Cell when their internal ID's match.
Each Cell and ProxyCell however has a unique name (not same as ID).
A Cell can reference (by name) a ProxyCell (as long as the ProxyCell does not represent that actual Cell - which would effectively be a self-reference and is not allowed).
What I want to do is find out where I have a reference from any instance of Cell - cell1 - to any instance of ProxyCell - proxycell2 - but am missing a reference from cell2 to proxycell1 where proxycell2 is a representation of cell2 and proxycell1 is a representation of cell1.
Here is the rule I have written for it:
rule "Check consistent references"
when
$cell1 : Cell()
$cell2 : Cell()
$proxycell1 : ProxyCell ( id = $cell1.id, $cell2.references contains this.name )
$proxycell2 : ProxyCell ( id = $cell2.id, $cell1.references not contains this.name )
then
//report an error.....
end
I have 10,000 instances of Cell and 10,000 instances of ProxyCell in working memory. For each instance of Cell, it can have references to 60 different ProxyCell instances. Loading the Cell and ProxyCell instances into working memory is not a problem. However, when I try to run this rule above, the memory quickly goes above 1GB and I eventually get an out of memory error.
I was wondering if there is a better way to structure or write this rule so that it doesn't use so much memory.
Thanks,
Ryan.
15 years, 11 months
Rules doubt
by Puneet duggal
@ friends,
i am facing a problem in my business rules and it is :-
when i make single mistake in my business rule rest all rules
stops working ,why is it so ? and how can i detect which exact business
rule is having
error
thanks
Puneet
15 years, 11 months
Drools Flow: Workitems with the same name
by tolitius
Hey Drools Crew,
Business process can be designed to have more the one workitem ( human
task ) with the same name. So two questions here:
1. Confirming: "When registering handlers all of the workitems with
the same name will use the same ( registered to the name ) handler?"
2. What is the way for the client to identify the workitem in this
case? Let's say there is a "truly" parallel definition:
/---> workitem1 ( async send to external
system, waiting for completion )
split ----> workitem2 ( async send to external
system, waiting for completion )
\---> workitem3 ( async send to external
system, waiting for completion )
If workitem1 and workitem3 have the same name, and the response
came back for workitem3, how do the client know to "complete" workitem3 vs.
workitem1?
Workitem IDs are completely workflow / drools-flow related, and
when ( let's say ) approval has returned from an external system, a client
should just tell one of the business workflow services to:
completeApprovalFor( ExternalSystem.SuchAndSuch )
workflow service will have a handler to a workflowInstance
{ businessId, sessionId, processInstanceId, List<Work items in progress>,
etc.. }, so if it is a single workitem the process is waiting for, there is
no problem, but if there are more workitems that make the process to wait +
they may have the same name, it creates a problem.
One solution (for the above specific case) would be to link external
systems with work item ids, so instead of List< Workitems in progress >, it
would be Map< ExternalSystem, Workitem >. But it just looks to complex and
specific...
Anything the framework can help with?
Thank you,
/Anatoly
--
View this message in context: http://drools-java-rules-engine.46999.n3.nabble.com/Drools-Flow-Workitems...
Sent from the Drools - User mailing list archive at Nabble.com.
15 years, 11 months
addKnowledgePackages Delay
by malkhafaji
Hello,
I am trying to add knowledge packages to an existing KnowledgeBase. My
largest DRL file has 110 rules. I noticed that it may take up to 5 seconds
just to execute the following statement:
this.knowledgeBase.addKnowledgePackages(packages);
where "packages" is just a list of one knowledge package representing one
drl. The nature of my rules is that I initially load a small set of rules
and then at runtime my rules may load other DRLs. The problem is each DRL
loading command (this.knowledgeBase.addKnowledgePackages(packages)) takes a
few seconds and I have no idea why. Although, all those DRLs are already
compiled (at startup time). It is the line I pasted above that takes the
longest time out of anything else I do.
Any idea why just merely adding a KnowledgePackage list to a KnowledgeBase
can take this much time? Do you guys have any suggestions on how I can
reduce this time? I tried to create another dummy KnowledgeBase and get the
KnowledgePackage from it at runtime and add that to my KnowedgeBase above
(thinking that would reduce the time maybe), but not at all.
Any ideas would be great. The API does not have much info on what that add
method exactly does behind the scenes, and how I can get around the long
delays. This is very frustrating to me, and I am not sure where to look.
Thanks for ANY help at all.
--
View this message in context: http://drools-java-rules-engine.46999.n3.nabble.com/addKnowledgePackages-...
Sent from the Drools - User mailing list archive at Nabble.com.
15 years, 11 months
Flow - handling multiple events of the same kind
by Ed Staub
If the same event-node is triggered twice with different data, the same
variable will be recycled for the different data payloads. How does one
ensure that data is not lost?
My best guess is to follow the event node with a composite node, and copy
the top-level variables into the composite's variable scope in the first
sub-node of the composite. But I'm not sure whether I can guarantee that
the composite will be able to do that before the engine handles the second
event.
Suggestions?
--
View this message in context: http://drools-java-rules-engine.46999.n3.nabble.com/Flow-handling-multipl...
Sent from the Drools - User mailing list archive at Nabble.com.
15 years, 11 months
