Python Server

[RichB]

Hi all,

I am gathering specifications for a server to run a python Server Side Extension. The goal is to set up a virtual machine to implement Machine learning/BI solutions. We currently run Qlik Sense Enterprise on seperate development and productions servers.

I am trying to choose between Redhat Enterprise Linux and Windows Server 2012. From the resources I have seen such as https://github.com/nabeel-oz/qlik-py-tools#usage windows is being used.

I am wondering if there is any reason I would not be able to run the SSE on redhat. What are the tradeoffs or drawbacks of each?

Thanks in advance!

[Nabeel_Asif · Accepted solution]

Hi RichB,

I'm the author of the GitHub project you mentioned. The project does have installation options for both Windows and Linux. Windows got the lengthier documentation as most Qlik users are more familiar with it, but that is not an endorsement for going down the Windows route.

I have used containerization to the run the SSE on Linux and seen much better performance than a similar sized Windows Server 2012 VM. Some Python packages are built for Linux and then ported to Windows so that is a factor in the performance difference.

IMO, the benefits of containerization alone make Linux a good choice.

There's a Docker image for my SSE available if you want to do your own experiment: https://github.com/nabeel-oz/qlik-py-tools#docker-image

[Tee_dubs]

Hi Nabeel,

First of all. Your project is awesome. I've been using Pytools Prophet on my notebook to create a PoC for a forecasting app. Im probably there target audience you were looking for. Im not a python expert, yet I want the capabilities it allows.

The PoC is running great and we are standing up a windows server to put it into production.

What server specs would you recommend to start with?  As I said, the first use case is Prophet on a 4 year hourly time series (multi-regression). I expect to find al lot more use cases though.

Thanks