Skip to main content


Eclipse Community Forums
Forum Search:

Search      Help    Register    Login    Home
Home » Archived » BIRT » performance drop
performance drop [message #1241097] Fri, 07 February 2014 13:34 Go to next message
osuwariboy Mising name is currently offline osuwariboy Mising nameFriend
Messages: 24
Registered: December 2010
Junior Member
Hello everyone,

I'm experiencing some very strange performance drop with a few arrays. The arrays themselves are as basic as you can get. They just display columns with the total at the bottom.

The problem here is that when I reach a certain number of columns, the performance just drops brutally. For example, I have table A with 19 columns. It takes over 1 minute to process 1000 records (i.e. Insanely slow considering the query is executed in less than 1 second). Drop the number of columns to 13 and the performance jacks up to 3 seconds. However, when I add just one more column, the performance drops back down to 1 minute.

I did some tests with various combinations of columns and I've concluded that it's not one specific column that causes the drop in performance, but the number of columns itself.

I had the same problem with another array, though for this one, the performance drop happened at the 11th column.

Anyone has any idea what could be causing this? I mean the arrays themselves are nothing special, it's just a set of columns formatted as currency with the totals at the bottom of the report.

Thanks in advance,

Osu
Re: performance drop [message #1241111 is a reply to message #1241097] Fri, 07 February 2014 13:57 Go to previous messageGo to next message
donino donino is currently offline donino doninoFriend
Messages: 183
Registered: July 2011
Senior Member
It looks like a heap memory issue: beyond some point your Eclipse JVM needs to run garbage collectors during execution. This would highly depend on which report element is used (cross-table, simple table...) and which settings (in particular "page break by" values)

It would be great if you could reproduce this with the sample database, and join the resulting report. Otherwise i think it will be difficult to help because it might be due to something specific to your context, or specific to this report.

Re: performance drop [message #1241126 is a reply to message #1241111] Fri, 07 February 2014 14:17 Go to previous messageGo to next message
osuwariboy Mising name is currently offline osuwariboy Mising nameFriend
Messages: 24
Registered: December 2010
Junior Member
If you're talking about the "classic model" database, the tables only have a couple hundreds lines at most, which is not enough to generate the bug. Is there one which has "big data" in it?
Re: performance drop [message #1241310 is a reply to message #1241126] Fri, 07 February 2014 19:54 Go to previous message
Michael Williams is currently offline Michael WilliamsFriend
Messages: 1925
Registered: July 2009
Senior Member

You could probably create a scripted dataSet with a loop for how every many rows you want and just use a single static row of data to reproduce that many times to be able to have more data to use.

Michael

Developer Evangelist, Silanis
Previous Topic:BIRT (EPL) licensing
Next Topic:Report Titles Using Plain Text or Auto
Goto Forum:
  


Current Time: Fri Apr 26 08:05:13 GMT 2024

Powered by FUDForum. Page generated in 0.02603 seconds
.:: Contact :: Home ::.

Powered by: FUDforum 3.0.2.
Copyright ©2001-2010 FUDforum Bulletin Board Software

Back to the top