Hi!
Since it is not possible to post replies in the Announcements section, I decided to go an and create a discussion thread for this announcement here:
First off, Happy New Year everyone! It's nice to see how Dathorn is getting better and better over time, especially, given that the last year was by no means easy and the years to come probably are not going to be any easier either.
Miam-miam-miam!
Other than that, I actually wonder if the new LiteSpeed infrastructure is going to make it any easier to support custom FCGI / SCGI / UWSGI backends in the future. Not that I am not happy with what I can do now, but using plain CGI clearly prevents anything like Python / Ruby to be used for anything serious other than just playing around...
Also, is RHEL6-based up-to-date Python / Ruby stack coming out-of-the box?
I wonder if you did any 32/PAE vs. 64 benchmarks. My hunch would be that the syscall overhead would kill basically kill the benefits of having untranslated memory for shared scenario when there are lots of small processes none of which actually needs / allowed to take that much, so probably 32 is still the way to go?
Since it is not possible to post replies in the Announcements section, I decided to go an and create a discussion thread for this announcement here:
First off, Happy New Year everyone! It's nice to see how Dathorn is getting better and better over time, especially, given that the last year was by no means easy and the years to come probably are not going to be any easier either.
Once these upgrades are complete you can expect a significant increase in our hosting plan disk quotas across the board.
Other than that, I actually wonder if the new LiteSpeed infrastructure is going to make it any easier to support custom FCGI / SCGI / UWSGI backends in the future. Not that I am not happy with what I can do now, but using plain CGI clearly prevents anything like Python / Ruby to be used for anything serious other than just playing around...
Also, is RHEL6-based up-to-date Python / Ruby stack coming out-of-the box?
I wonder if you did any 32/PAE vs. 64 benchmarks. My hunch would be that the syscall overhead would kill basically kill the benefits of having untranslated memory for shared scenario when there are lots of small processes none of which actually needs / allowed to take that much, so probably 32 is still the way to go?
Comment