I wrote about this subject recently:

http://www.thisdev.com/2009/11/i-want-to-like-microsoft-azure.html

but the problem isn't just about small sites. The whole instance-based scaling model is wrong. It makes Azure a limited hosting environment instead of a powerful, scalable cloud service. One of the other suggestions is "Provide multiple roles per instance" -- this makes no sense if Azure just scaled properly, automatically. Same goes for "Make it easier for me to scale up or down my Windows Azure hosted service", "Provide me with flexible pricing models...", and probably others as well.

As my blog post says, if you don't fix it, I don't know why anybody would use Azure over the competition (GAE and EC2).

Well, I think Rob G pretty much nailed one side of the story. Kicking the tires and trying to grow small fish into big ones is one scenario. Serving small companies or even consumers as a web page hosting platform is another. They're never going to grow, but there are ridiculously many of them.

Having an app-level cloud abstraction (as opposed to cloud-hosting a normally managed virtual machine) would be a great way to bring down the maintenance cost of small ASP.NET sites. Whether or not Azure should be competitive in that segment is another question, but definitely the Microsoft stack is at a disadvantage here.

Hi Mike,

I appreciate you mentioning the introductory offer but with all due respect, that offer is not even on the same planet as the competition. In the interests of full disclosure, I already am a Microsoft Partner and I have the MSDN subscription that nets me some extra hours, but I'm not convinced yet. I have a small start-up that I'm bootstrapping with my own cash. How attractive do you think Azure looks to me given the current pricing model?

You could be blowing the competition out of the water, but you've come to the table with fairly weak pricing and an even weaker introductory offer. Just take a look at the small instance allowance. 25 hours - that's little over a day that I've got the instance turned on, never mind the fact that zero traffic is coming my way yet - which is the very thing that allows me to determine if the business and Azure are viable choices for me. Kicking the tires needs more time or bandwidth, and you're not offering that. What do I do for the other 30 days in the month - pay for no traffic? Sure I'm happy to - but at a much lower price. Something needs to be attached to actual usage in terms of traffic, not just the green light powering the machine.

I also disagree with Dennis I'm afraid. I think that Azure should be trying to attract exactly that $10/month business! It's the small guys that get big remember, the big fish already have their own data centres and infrastructure set up, and while I appreciate you'll be working hard to land large accounts, hundreds of thousands of smaller accounts amounts to the same thing and it is those guys you are forcing to the curb and the alternative is Google (on the same model) with a very generous month-on-month quota, and Amazon is catching on fast by giving plenty of free SimpleDB traffic away every month before charging.

It's pricing strategies like these that help smaller guys get started, and enable them to get bigger, which in turn with net you more money of course. Once I've linked into the Google or Amazon APIs, there's not really a very big chance I'm coming back out - let's be honest about that, but right now you have a very good chance syphoning off some of that interest in your direction.

I hope this hasn't come across as all negative, I am very excited about the possibilities Azure has to offer. My solution happens to be a Asp.Net MVC 2 application and the two go hand in hand and I very much hope things work out on Azure. Time and pricing will tell, but right now, I can only see how much cash this is going to cost before it's even up and running.

Kind regards,
Rob
http://blog.cozwecan.com

Yep, definitely. By my rough calculation it would cost around £1000/year to run a simple WordPress blog on Azure - you've got to have some pretty effective adverts running to justify that!

The problem is that compute hours don't take into account usage, so you could have an instance sitting idle for a month and it would cost exactly the same as a highly active instance.

One of the big potential benefits the cloud can offer is the ability for developers to test the waters with a service to see whether it takes off, without needing to invest in infrastructure upfront - if users like it and it becomes the next Facebook, you can grow quickly (and pay accordingly) without disruption to service. If it doesn't work out, you can scrap the whole thing without having wasted money on hardware you never needed.

Google AppEngine and its free quotas makes this a reality; Azure on the other end has too high a barrier of entry - there's no way I'm going to experiment with the next Facebook on Azure when it's going to cost me so much. Developers are therefore likely to try out AppEngine and, if they find their service successful, are almost certainly going to stick with (and start paying) Google rather than switching to Azure.