Search code examples
pdfpdf-generationcognoscognos-10

Cognos 10.2.2 PDF File Size / Manual vs Burst

I have an IBM Cognos report that runs everyday on burst mode and delivers the contents to ~60 people. Each person gets his/her own set of data.

They are all around 300KB PDF files, except for one of them that is a 2.5MB file. It's not the first nor the last generated report, if that matters.

The problem is, when this same person's report is generated manually, I get a small (300KB) file. Every burst output gives me a 2.5MBish file, for this person only!

I haven't found anything like this online, either. Some information I could extract from the PDFs are:

Manual:

Filename: Manual.pdf
Producer: PDF Engine win32 - (10.2)
Created: Thu May 26 09:02:01 EDT 2016
Page count: 21
PDF version: 1.4
Linearized: version 1
image: object 57, 1200x22, raw samples, depth = 8, colorspace = DeviceRGB, length = 99
image: object 58, 1200x22, raw samples, depth = 8, colorspace = DeviceGray, length = 338
image: object 61, 144x72, DCTDecode, depth = 8, colorspace = DeviceRGB, length = 4953
image: object 65, 1200x25, raw samples, depth = 8, colorspace = DeviceRGB, length = 110
image: object 66, 1200x25, raw samples, depth = 8, colorspace = DeviceGray, length = 110
FONT NAME                        TYPE          ENCODING      EMB SUB UNI   OBJ# 
Arial-BoldMT                     TrueType      WinAnsi        Y   Y   N      46
Arial-ItalicMT                   TrueType      WinAnsi        Y   Y   N      52
ArialMT                          TrueType      WinAnsi        Y   Y   N      49
69 objects
    3 fonts:  3 TrueType   3 embedded   [AMEMFV+Arial-BoldMT, EUJZHO+ArialMT, TQJQWB+Arial-ItalicMT]
 (21 pages)  Content streams command usage:   w 11459   J 11459   d 11459   q 12168   Q 12168   cm 421   m 23082   l 23082   re 625   S 23110   f 288   n 309   W 309   BT 11623   ET 11623   Tf 11623   Td 11623   TJ 11623   RG 11459   rg 11911   Do 400

Burst:

Filename: Burst.pdf
Title: OBFUSCATED
Author: Reports, US
Producer: PDF Engine win32 - (10.2)
Created: Thu May 26 06:26:18 EDT 2016
Page count: 21
PDF version: 1.4
structured
Linearized: version 1
image: object 25539, 1200x22, raw samples, depth = 8, colorspace = DeviceRGB, length = 99
image: object 25540, 1200x22, raw samples, depth = 8, colorspace = DeviceGray, length = 338
image: object 25543, 144x72, DCTDecode, depth = 8, colorspace = DeviceRGB, length = 4953
image: object 25547, 1200x25, raw samples, depth = 8, colorspace = DeviceRGB, length = 110
image: object 25548, 1200x25, raw samples, depth = 8, colorspace = DeviceGray, length = 110
FONT NAME                        TYPE          ENCODING      EMB SUB UNI   OBJ# 
Arial-BoldMT                     TrueType      WinAnsi        Y   Y   N   25528
Arial-ItalicMT                   TrueType      WinAnsi        Y   Y   N   25534
ArialMT                          TrueType      WinAnsi        Y   Y   N   25531
25551 objects
    3 fonts:  3 TrueType   3 embedded   [AMEMFV+Arial-BoldMT, EUJZHO+ArialMT, TQJQWB+Arial-ItalicMT]
 (21 pages)  Content streams command usage:   w 11459   J 11459   d 11459   q 12168   Q 12168   cm 421   m 23082   l 23082   re 625   S 23110   f 288   n 309   W 309   BT 11623   ET 11623   Tf 11623   Td 11623   TJ 11623   RG 11459   rg 11911   Do 400   BDC 11912   EMC 11912

Some key differences I'd like to point:

  1. Burst report has that "structured" line
  2. Burst report has 25551 objects against 69 for Manual
  3. I've extracted fonts and images, and they match bitwise

Every help is highly appreciated!

Solution

  • Turns out that the server had different compression settings for ReportJob and BatchReportJob. I am still not sure on why only one of the 70+ reports had a huge size, but tweaking those settings fixed the issue.