Perhaps I wasn’t clear enough. With the older RoboHelp model you are limited to one physical machine, which must be a Microsoft IIS machine, for all content hosting and reporting functions. In the MadCap model the repporting engine requires installation on a Microsoft IIS machine, but it can be one little machine in the corner while your actual documentation content can be on any single machine of any “flavor” or on any kind of server farm. The connection between the published content (say, on a farm of 10 Apache machines) and the MadCap Feedback Server (running on a single Microsoft IIS machine) is virtual. This way for the load sensitive pages (the actual content) you have all of the flexibility necessary. Then for the non-traffic sensitive component (the tracking and Web 2.0 functionality) the MadCap Feedback Server only needs the single IIS machine install.

I hope that helps,

Mike

“In a MadCap Software tool work flow generated content files can be run on the same Microsoft IIS server as the MadCap Feedback Server, or your content files can be hosted on a single Unix or Linux machine or even on a farm of machines for load balancing. This can be critical for high traffic situations. It also comes in handy if your company is a Linux/Unix shop, you won’t have to explain why you need a bunch of new Microsoft servers just to host your documentation.”

From the system requirements:
http://kb.madcapsoftware.com/default_Left.htm#StartTopic=Content/Feedback/Installation/IAR1002FB_FeedBack_Server_Req.htm|SkinName=MadCap

Is this contradictory? Or do I just not know enough to understand that it makes perfect sense?

Great summary of points! Of course this is right up my alley as far as research for the past year has shown…