Github - TDWTF-Forums

skotl

I'd agree with that. And we don't have real-world tools to evaluate how much virtual pixie dust we need for each application.
I love the idea of elastic-scaling but I'm petrified of the idea of elastic-billing, and how far that piece of elastic might stretch.

Matches

A lot of times your available throughput is based directly on your allocated disk space, since iops typically get assigned by it. Azure has their own terminology, and i work with aws because it's considerably cheaper.

skotl

s'my point - it's pretty much impossible to predict your requirement and, more important, it's impossible to predict the behaviour if you undershoot.

tar

If this is useful, this is (more or less) the list of .NET assemblies I'm currently using for webserver-type stuff:

httplib/contrib/server at master · stdjon/httplib

HTTP Webserver written in Nemerle. Contribute to stdjon/httplib development by creating an account on GitHub.

If it's not useful, know that at least I tried.

EDIT: wow, that was a useless onebox...

Matches

That's not what I said at all. You can predict expected throughput requirements easily enough if you know about your application and your hosting platform. I can tell you exactly what's needed for physical hardware or AWS. I'm just not interested in learning azure because across the board it's more expensive than aws.

blakeyrat

@Matches said:

I'm just not interested in learning azure because across the board it's more expensive than aws.

I know a guy using Azure, and he says it can be cheaper if and only if you rewrite your app for Azure. (Using the Azure WebApp and Azure WebJob templates in Visual Studio.) The trade-off is: that basically locks you in to only using Azure.

If you're doing VMs, AWS will always be cheaper.