Stephen De Gabrielle wrote on 11/01/2017 04:46 AM:
I *suspect* the whole ‘scaling Racket’ thing is a furfy, and for the
vast majority of applications or websites this is never going to be a
problem. I don’t *know* if this is true, so I’d welcome some _evidence_.
I agree that some of the industry chatter about "scalability" is just
chatter. And it's a good grown-up-sounding term to use when we don't
know anything else to say about a platform. But I think the reason it's
grown-up-sounding is that, at one point, our honorable ancestors learned
that servers have performance characteristics and requirements, that
naive approaches are often simply inadequate, and that "scalability" is
a real concern.
Two scenarios of being concerned about scalability:
* If my new-college-grad startup team does a bunch of coding like the
platform tutorials say to do, and we don't yet understand much of what's
going on, will the result perform well enough on an EC2 large instance
for the moderate-volume webservice of our iPhone app, despite much of
our own startup code being sloppy? (Answer: Maybe...) And will we be
able to keep it working acceptably if our beta users grow
ten-thousand-fold literally overnight? (Answer: We could probably
simulate that load, against our evolving prototype, in a testbed server
setup.)
* If my highly-skilled team uses this platform in our architecture, when
we need our processing to handle X requests per second with Y
responsiveness, and we know we'll have to be creative regardless of
which platform we use, is this platform's part of it going to perform at
least as well as if we'd used this other platform for that part, or are
we going to have to work harder or pay more for hosting to compensate?
(Answer: Maybe...) And are there platforms that are more proven in this
regard? (Answer: Yes.)
I suggest... We appreciate that scalability of one's use of a platform
is a legitimate concern, and hard to answer, and the "platform" is just
one factor. And we know that, with Racket, like most other platforms, a
CS 101 student can't just make a poo in an IDE and call it a high-volume
Web site. Usually. And we don't assume that we scale in ways that have
never been proven. And we know people feel better about picking a
platform as a part of their architecture when they see successes that
managed to use that platform in a similar way.
For your evidence, I think the most-known public server deployment among
developers is probably Hacker News. You can ask them how their setup
works currently, and how much of Racket is being used, and what they had
to do around that. (I suspect their answer is not that different than
my own production server experience: most of it is not what current
Racket docs imply is the platform, there was also some standard
extra-platform infrastructure, and clever engineering was required on
some tricky parts.)
What I suspect Racket industry promotion wants is a startup team that
does their rapid prototype in Racket, and, when it comes time to move
forward, they are technically strong enough to assess things and see a
way that they can confidently keep using Racket. (Good news: clever
engineering, elbow grease, and cloud servers make most things possible.)
--
You received this message because you are subscribed to the Google Groups "Racket
Developers" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
To post to this group, send email to [email protected].
To view this discussion on the web visit
https://groups.google.com/d/msgid/racket-dev/cf5c8857-3a00-ff64-d77e-f8cc2806a791%40neilvandyke.org.
For more options, visit https://groups.google.com/d/optout.
