Hacker News new | past | comments | ask | show | jobs | submit login

Given Google's reputation of awful customer support, why would anybody want to locked into their platform, especially when you have to do so many things the "app engine" way?



You can run Django on top of AppEngine and port your app to any number of Django hosts with minimal changes. I just talked to a startup founder who started out on AppEngine and just recently switched to Heroku.


Last year I spent a number of months porting an (albiet poorly written) django app from AppEngine to AWS. I would not call the number of changes we made in order to port the app "minimal". I have heard similar stories from other development teams. My evidence is totally anecdotal, but so far it's lead me to believe AppEngine is not a good solution for running a django app.


Did you happen to ask why they decided to move from AppEngine to Heroku?


Didn't offhand - I just saw him in the hallway for 5 minutes or so while he was in town for I/O.


They have introduced upgraded "Support Packages" for a premium: https://cloud.google.com/support/packages


If you have an application that happens to fit nicely into the App Engine Way, it's a pretty slick environment for deployment.


Yes, I have no argument with you that it is quite slick. It seems the benefit still does not outweigh the cost though. I can't even imagine running into a critical issue with paying customers demanding answers from me while I am subjected to Google's customer service. Plus there's no way to failover from App Engine to some other platform if your app is locked into the "App Engine Way".

The only thing I would consider using App Engine for at this point would be a school project or something.


> Plus there's no way to failover from App Engine to some other platform if your app is locked into the "App Engine Way".

Well, except that there is an open-source alternative implementation of App Engine that you can run anywhere you want. [1] Which kind of ruins the whole "no way to failover from App Engine to some other platform" argument, even for apps built the "App Engine Way".

[1] http://www.appscale.com/


> Well, except that there is an open-source alternative implementation of App Engine that you can run anywhere you want.

To be fair, AppScale is a separate entity and therefore there's no guarantee that they will keep up with GAE API's, API bugs, and/or quirks. These are obviously things out of their control, but I'm saying this because there's still no guarantee of a 1:1 replacement if you need to move away.


It's an incredibly quick way of getting something out there.

If agility matters, I've not found anything that beats it. You might not want your service running on there as you scale, but it's enough to get a few thousand users onboarded as a proof of concept.


Some parts of app engine are very much designed to be really scalable and geographically redundant. Take sharded counters, for example [1], if you want to count something that happens more than 5 times a second.

If you want a proof of concept and you're not concerned about getting the scalability, why not run on a single server?

[1] https://developers.google.com/appengine/articles/sharding_co...




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: