We waited a long time for Microsoft to finally announce their Azure pricing as they have on their blog and an unveiling with much fanfare at their partner conference keynote as a type this. Frankly, after waiting for so long, I expected Microsoft to come up with something a bit more imaginative than a copy/paste of the Amazon EC2 pricing. Yes, they promised to ‘be competitive’ and they are – a compute hour on Azure will cost $0.12 and the Amazon one is $0.125; excuse me for not thinking that that is very competitive or even remotely interesting.
Months ago I was at an Azure technical briefing where, during the break, some of us were speculating as to how Azure would be priced and how different pricing options would encourage system behaviour, and therefore design, that would be cost optimal instead of performant (or meeting requirements even). High pricing of bandwidth would encourage higher use of CPU for compression – and visa versa. It doesn’t take long to speculate on all the possible metrics and how you could price around those metrics and, in turn, design for them.
It would seem that the marketing folk have won out over the engineers… I can picture meetings where MS engineers argued all sorts of interesting metrics that could be measured and the data centre people weighed in with their costs models and after all the back and forth some pointy haired manager from marketing said something like “Let’s make it easy for the market to understand” and, while everyone waits for the insightful nugget he leans back in his chair and says “Someone pass me that EC2 pricing… I want to have a look at what it says”.
Microsoft had an opportunity to really take a leadership role in the cloud computing business and create a pricing model that makes people finally sit up and pay attention to the cloud. They own virtually the entire stack – from .NET to Azure Fabric, to the web server down to the OS that it all runs on (including the VM network card drivers), so I am sure they could find some place to pick out some interesting metrics. Instead they have come up with something that, like the tea in the Hitchhiker’s Guide to the Galaxy (which tastes almost, but not quite, like tea) is priced almost, but not quite, the same as EC2 or anything else that may be out there.
The hosting that I use for my own personal projects is from discountasp.net which gives me a whole lot of stuff for $10 per month and another $10 if I want SQL Server. Is it good enough to host a small business? I think so, but I am not sure as I have never really tried it. Microsoft has been positioning Azure to smaller businesses as the right solution (low initial cost and ability to scale) so their competitors are not necessarily Amazon, but also traditional multitenant hosters such as discountasp and more managed hosting offerings from the likes of rackspace for when the business starts to grow.
The opportunity that I think that Microsoft has lost is the low barrier to entry for small businesses. Azure starts at $100/month and throws in vendor lock-in for free, which may not be compelling for someone who has the choice to register a domain name and have a .net website up for $100 for the whole year. Microsoft should have made Azure virtually free say $5/month or $10/month for three websites in order to attract people to the platform, lock them in and hope that they grow. Instead they have pricing options that, when assessed with all the competitors out there, are not particularly attractive. Also, while the storage transactions and .NET service message based pricing is interesting it sounds a lot like ‘transaction tax’ – so you are paying for compute hours, plus bandwidth, plus (transaction) tax. Again, when assessed against competitor offerings, doesn’t sound that great.
While much discussion will ensue on the Azure pricing and (yet to be detailed) SLA’s I can’t help shaking the feeling of disappointment. It’s not that I expect everything to be free (although Google AppEngine has a free threshold), but I expected something compelling in terms of innovative pricing as well as something sweet to attract businesses and developers to the platform – not just a rehash of an existing pricing model. Surely Microsoft’s business is different from Amazon and they could come up with something that better reflects their unique position (and ambitions) in the market?
Granted, this blog post has been hammered out in the minutes after the announcement was made and before the official presentation at the partner conference, so maybe I don’t have the complete picture – if my views change over the next few days I’ll update them here.
Simon Munro
@simonmunro