I have done no load-testing of either BlueDragon or Railo in fact I have never worked formally with a client who used either and I have been working on site with clients since late 1999.  That to me is the most telling issue for both OpenBD and Railo to surmount if they want to succeed on a large scale.  For better or worse, CFML is not viewed in the same light as PHP, Ruby Perl etc.

In addition I believe that confusion about licensing, as evidenced here (in the CFGuru thread), will hurt OpenBD and Railo and that needs to be clearly defined in terms business users can understand.  The difference in cost between ColdFusion, OpenBD and Railo even if OpenBD and Railo are totally free is largely irrelevant in the enterprise world if confusion over licensing exists, that is my opinion.

Also, differences in tags-functions is still a cogent issue and will remain so and obviously the CFML committee was never allowed to flourish at all.

Lastly, I have always been able to get ColdFusion to fly, with proper tuning-DB massaging etc, I expect Mark Kruger can attest to that too.  To be fair, much of the slowness I have seen in CF applications is caused by outside items, databases in particular. One caveat here is CFC's they really have brought poor performance to ColdFusion in my experience and the more heavily they are used the worse that gets.

I wish both OpenBD and Railo great success, more CFML applications are good for all of us in the community.  Perhaps, they should push heavily into Cloud based architectures?