Skip to Content.
Sympa Menu

grouper-users - RE: GrouperClient GcGetGrouperPrivilegesLite timings

Subject: Grouper Users - Open Discussion List

List archive

RE: GrouperClient GcGetGrouperPrivilegesLite timings


Chronological Thread 
  • From: Chris Hyzer <>
  • To: Paul Gazda <>, Grouper Users Mailing List <>
  • Subject: RE: GrouperClient GcGetGrouperPrivilegesLite timings
  • Date: Fri, 26 Jun 2009 16:42:29 -0400
  • Accept-language: en-US
  • Acceptlanguage: en-US

Sorry, you need to run the server side in yourkit, not the client.  Fire off the client (not in yourkit), and hopefully we will learn something.

 

Thanks!

Chris

 

From: Paul Gazda [mailto:]
Sent: Friday, June 26, 2009 4:40 PM
To: Chris Hyzer; Grouper Users Mailing List
Subject: RE: GrouperClient GcGetGrouperPrivilegesLite timings

 

I re-ran the gcGetGrouperPrivilegesLite.execute() tests using YourKit. It doesn’t seem to provide any insights. The time difference is in the java.io.

 

CPU times when using ActAsSubject WITHOUT source specified:

 

CPU times when using ActAsSubject WITH source specified:

 

 

Paul Gazda

 

From: Chris Hyzer [mailto:]
Sent: Tuesday, June 23, 2009 4:23 AM
To: Paul Gazda; Grouper Users Mailing List
Subject: RE: GrouperClient GcGetGrouperPrivilegesLite timings

 

First of all, 2600 millis seems slow in general for a call.

500 millis for any subject lookup is a lot, but a 500 mili difference between specifying a source, and not specifying a source, is also a lot, and yes, it doesn’t really make sense that it would go in that direction.


Here is the code which resolves the actAs subject:

 

        if (hasSubjectSource) {

          this.subject = SubjectFinder.getSource(this.subjectSourceId).getSubject(

              this.subjectId);

          return;

        }

        this.subject = SubjectFinder.findById(this.subjectId);

 

Is there any way you can run the web services portion in a profiler like YourKit (you can get a free trial) to see which parts of the code are taking the time?

 

Since this is also a common thread, maybe you could try the LDAP pooling source adapter that was recently posted to the list… 

 

Chris

 

From: Paul Gazda [mailto:]
Sent: Friday, June 19, 2009 12:47 PM
To: Grouper Users Mailing List
Subject: [grouper-users] GrouperClient GcGetGrouperPrivilegesLite timings

 

I have been trying to make my calls using GrouperClient as efficient as possible by running some tests and monitoring the values in getResponseMetadata().getMillis(). When I ran some tests using gcGetGrouperPrivilegesLite.assignActAsSubject, I found something that seems counter-intuitive and just wondered if anyone can explain the reason for the timings I am seeing.

 

When I instantiate the WsSubjectLookup object used by gcGetGrouperPrivilegesLite.assignActAsSubject with a subjectSource, gcGetGrouperPrivilegesLite.execute() runs 20% slower than when I define the subjectSource as null (average of 3175 millis compared with 2641 millis). In this case, the subject source is LDAP.

 

I would have thought that defining the subjectSource would make whatever subject lookups are being done run faster rather than slower, because the source is known. I’m also wondering whether this kind of performance difference would always occur when WsSubjectLookup objects are used (e.g. gcGetGrouperPrivilegesLite.assignSubjectLookup ).

 

Paul Gazda




Archive powered by MHonArc 2.6.16.

Top of Page