Performance of ACIF vs PDF Indexer

[kurtschwanz]

We ran some tests yesterday and were surprised shocked by the results - wondering if anyone can explain the gap?

We took an AFP file that contained 13K+ documents (65k+ pages) and converted it to a PDF with PPDs.   

We loaded both documents into identical applications (except the indexer, obviously).  The ACIF indexer took two minutes. The PDF indexer took 45! And it pegged the server CPU the whole time.   We thought perhaps this was because the PDF had compression enabled during the conversion, so we converted it again without compression....that one took 51 minutes to load and again killed the CPU for a large portion of that time.

We fully expected there to be some lag with the PDF, but not to this degree.   Is this just the way it is, or are there perhaps some tweaks we missed that would bring this more in line with the AFP?   I think we all expected it would take no more than 10 minutes to load.

OnDemand 9.5.06 running on Windows

[Nolan]

PPDs are supposed to be as fast as gets for PDF documents.   Can you share the all PDF Indexing message?

[kurtschwanz]

Attached is the output from the load of the compressed PDF.  Let me know if you want to see the uncompressed also. 

[Justin Derrick]

I can't answer your performance issue, but I believe you should enable compression if you're using the PDF indexer with PPDs.

Also, double check your PDF -- 92MB of resources is crazy big.

I'd suspect there's a problem with your converted PDF file -- both for the resource size, and the duration it takes to process it.

The demos that IBM has given show that it's blazing fast -- so the performance problem you're seeing isn't typical.

-JD.