Performance Issues (CPU usage close to 100%)

Hello,

We are customers of GroupDocs, we bought an SITE OEM license (if you need more information please ask).

We have deployed GroupDocs Viewer (java) on production an we are experiencing some serious performance issues. I tested just for one user to view a pdf or a ppt file (attached) and the CPU goes close 100% just for one document. The documents are simple, 1-2 pages. See attached screenshot, on the spikes there I previewed a document with groupDocs.

Our application is deployed on EC2, and because of this issue we had to triple our EC2 machine fleet.

Please advise on configuration or how we should proceed since this is urgent for us (it kills our availability).

This is a document that we cannot preview in groupDocs http://www.pdf995.com/samples/pdf.pdf
but it is keeping the CPU at 99.9 for some timeā€¦

Document that works but CPU goes to 99%: Redirect Notice

Hello ,

Thank you for your inquiry.

Sorry, but from your request is not clear which a version of the GroupDocs.Viewer library you use exactly. Please specify it.

We have tested your files (pdf.pdf and Topic11SortingAndSearching.ppt) with 2.14.0 version and we don't see any serious performance issues. All files are rendered very fast (5-10 sec)

Also we glad to inform you that at the moment our product team works on the new generation of the GroupDocs.Viewer for Java library and this library also will have some performance improvements compared with 2.14.0 version.


---------

Best regards,
Evgen Efimov

http://groupdocs.com
Your Document Collaboration APIs
Follow us on LinkedIn, Twitter, Facebook and Google+

Thanks for your answer, we are using 2.14.0
The files are generated in 5-10sec, but do you see the spike in the CPU?

We have multiple clients, and when more requests of document view come to a server, it will be more than 10seconds at 100% CPU meaning that the server cannot serve other requests in that period of time. Can we configure somehow groupdocs not to take so much CPU?


Hello,

Our product team always works on performance improvements and they spend a lot of time for this part (optimization of the library and resources) . But unfortunately with current version of the library we can't reduce CPU usage .

Also, if you not use a pre-generation of the files cache in your integration, then you can try to implement it . It give you opportunity to generate cache for all files, before they will be called for view. Please investigate this article for how to implement CacheDataHandler.

Sorry for the inconvenience.

------

Best regards,
Evgen Efimov

http://groupdocs.com
Your Document Collaboration APIs
Follow us on LinkedIn, Twitter, Facebook and Google+