3 ms·
The problem with how Heroku handled RapGenius is that they basically have punted each time they have said almost anything with an apology and a promise for more
by programminggeek 14y ago
The problem with how Heroku handled RapGenius is that they basically have punted each time they have said almost anything with an apology and a promise for more info in the future.
What they probably should have said is that at some point if you are running too much stuff on Heroku, you probably should be running your own servers tuned to your own very specific workloads. Of course, that would sort of violate the whole premise of a PAAS in the cloud that can scale to infinity, but the point remains.
If you are paying $10,000+ a month, that buys you a lot of server time elsewhere that would probably be tuned better for current workloads or maybe Heroku needs to offer more turning options.
Either way, it feels an awful lot like Heroku has punted every time they had a chance to run with the ball on this and that's pretty disappointing.
- tlrobinson 14y ago"if you are running too much stuff on Heroku, you probably should be running your own servers tuned to your own very specific workloads" ...and if you're using a blocking, single-threaded server/framework your definition of "too much stuff" is a lot less than everyone else's.
- programminggeek 14y agoI agree, and honestly part of the problem is that people seem to have this notion that Heroku, AppEngine, or Azure will scale to infinity without any extra work on your part. At some point regardless of your tech, if you are at a large enough scale, you need to control your own destiny.