Free Rails hosting?

[Lothar Scholz]

Hello Doug,

DB> if you had a way to register what modules your app depended on, then you
DB> could make sure to load those in the master ruby process that the fcgi
DB> children would be forked from. copy-on-write would then make the kids
DB> use less memory than if each fcgi was spawned on the fly and loaded the
DB> modules on their own. that would be happening at a lower level than
DB> ruby's GC, so i figured that would work.

DB> but it sounds like you're saying that even if the code did that, the
DB> first GC run in the child would walk the list of data/code nodes and
DB> twiddle them in one way or another, causing copy-on-write to happen for
DB> all nodes, and negating any benefit from being forked from a master
DB> parent process. did i read you correctly? if so, that's interesting, i
DB> never expected that...

Yes seems that you got my point. A lot of the ruby data part will
always be copied with the first GC run. I don't know how much memory
is used by the interpreter and the *.so extensions, but i doubt that it
is not the huge part. Each data member has a flag as part of the
allocated memory that will be flagged in the mark phase.

But memory consumption is like performance profiling: not very
intuitive, so someone with a lot of time should measure the things
a little bit more. But i pay him a few beer's on the next Ruby Conf
if i'm wrong with my assumption.

DB> doug

 

[Kirk Haines]

that's pretty good performance, although it's difficult to assess
without knowing what your app is doing. this is why it will be great to
get some good, relative benchmarks between the different frameworks out
there.

Yep yep. Meaningful benchmarks can be hard.

i agree it's hard to do fair benchmarks, but i think it's definitely
doable. are you thinking just ruby and java? i would be glad to help
out with this, i have some background in performance testing.

I'm thinking primarily to look at Ruby approaches, but with rolling in a
comparison with some common Java standard approach, and perhaps a Perl
and/or Python standard approach, as well.


Kirk Haines