On Wed, 2010-05-05 at 16:02 +0000, Alex Clark wrote:
> As far as limitations of the platform goes, is this a "cloud computing"
> vs. "traditional model" issue? Or is it a GAE-specific thing. I suspect
> the latter because it seems any framework that runs on their (GAE's)
> platform has to bend itself to fit.
BFG doesn't really bend itself to GAE at all. There's not really any
feature of BFG that doesn't work in GAE. These days, even Chameleon
runs on GAE.
But BFG alone just doesn't do very much. It doesn't concern itself with
persistence or user data, etc, and all those pesky things that are
required for an actual application. So it's really apps written atop
BFG on GAE that need to bend themselves to GAE.
GAE allows for no data storage except via BigTable, no non-blessed C
code, requests need to finish in X seconds, the application has to start
very fast, etc. The limitations are existential: if they didn't exist,
there wouldn't really be a GAE, or at least "GAE" would be just another
system where you provisioned a machine, got a shell account and did
whatever you needed in a virtual OS environment, ala EC2.
In the meantime, the kinds of customers we (Agendaless) have are
typically larger organizations that have portability and futureproofing
concerns that aren't very well served by creating an application that
only works under a single vendor's environment. You can't take code
written for GAE and just deploy it somewhere else; you have to rewrite.
They also often have existing internal authentication databases and
other legacy systems that need to be integrated to. This just isn't
compatible with deploying to GAE.
So while GAE is cool and compelling technology for small personal apps
and startups who can afford to risk their own capital to bet that Google
will continue the service forever, it's not really a good fit for the
kinds of customers we usually get calling us on the phone.
> Whereas something like Ian's Silver Lining (which also excites me)
> might make a better "cloud deployment story" for BFG.
Repoze-dev mailing list