Tag Archives: Performance

Obiee 11g: After upgrade from 10g, 11g report generated UNION ALL in SQL

Applies to:

Business Intelligence Server Enterprise Edition – Version: 11.1.1.3.0 [1905] and later   [Release: 11g and later ]
Information in this document applies to any platform.

Symptoms

1) Customer upgraded 10.1.3.4.1 RPD and web catalog to 11.1.1.3.
2) Its noticed that in 11.1.1.3 the same reports take a lot longer to execute:
Eg: 1 min 30 sec in 11G vs. only 30 sec in 10G.
3) RPD doesn’t have any metadata inconsistencies, connection pool settings are good.
4) The memory usage spikes up when reports are run, and BI Server is taking too long compiling these reports.
5) Customer notices that the XML copied from 10g to 11g creates a more complicated logical query in 11g compared to 10g.
6) When customer copies the logical query from 10g and runs in 11g it gives the same performance as 10g.

Cause

Basically this is caused by many column with the empty string formula: ”, causing it to have 2nd UNION leaf to calculate the subtotal in 11.1.1.3.

Solution

These columns with a column formula ” where used as separators in the report.

Replaced all column separators with “duplicate measure” columns with a column length of 0.
This ensures the subtotals are NOT trying to use the separators as dimensions and create extra queries with UNION ALL.

Finally, it’s official – Firefox has better performance than IE

Browser performance is a hot topic. I normally develop with Firefox, but from my experience, it’s a rare situation that Firefox is supported by company’s IT department. Hence, I need to use IE on a regular basis – and I can’t say I’m too happy.

I feel like Captain Obvious for re-posting this, but here you go:

IE takes significantly more time than Firefox to open a report.

There are two issues:

1. MacAfee anti-virus software was configured to scan the OBIEE files. As a result, a scan was done on every file when, using IE, which significantly degraded performance. Firefox did not scans any of  the files, and therefore, was much faster.

2. Fundamental differences in the browsers. For example.. IE is built in as part of the OS, while Firefox is standalone.

To minimize the performance differential between IE and Firefox:

First of all, disable virus scanning on any Analytics (OBIEE) folders, and also, the web server folder where cache files are generated.

Secondly, Oracle does recommend any specific settings for either IE or Firefox, however, it is generally acknowledged that Firefox is a faster browser than IE. The explanation for this is that IE is embedded into the OS, whereas Firefox is a standalone application. The effects of this are:

  1. Memory Management: IE doesn’t do memory cleanup, because it is done by the OS. Firefox, however, only has to interact with the OS, where it does a better job of managing memory resources, first reserving, and then releasing memory to be reused faster.
  2. Network issues: IE being integrated into Windows uses the network settings from Windows, whereas Firefox can more efficiently figure out which settings to use.

Firefox eats IE in OBIEE