Some Windows Azure Thoughts

I’ve been pondering Windows Azure lately.  I know that a lot of folks are investigating the great potential of the cloud to offer on demand or even throw away computing resources.  Two things come to mind when I consider where this particular technology is going to take us.  I’ll demonstrate these two scenarios via examples:

1.  Windows Azure as a cloud based app for connecting to social networks.  http://webslice.cloudapp.net is a site from MS land that connects you to Facebook and pulls in information to the site and then spits it back out as a web slice for IE 8+.  The concept here is to leverage social media to do things, powerful .Net programming and hosting and a simple display – the web slice itself.  This web slice could be taken to several further conclusions – a consumable web part for SharePoint (not just for Facebook, for anything), a Windows Gadget, an endpoint for another application (like an aggregator that’s better than just a feed) and more.  There’s the added benefit of some wicked cool analysis that you could do on traffic coming in and out.

2.  Throw away computing resources.  I was talking to a colleague of mine the other day about an idea that I’m drumming up for the healthcare industry.  The idea is basically this – use Windows Azure for nightly batch processing and data storage.  All of the Windows Azure data centers either have or are getting their HIPPA/SOCKS/etc compliances so things are kosher there.  Next, take data that requires mountains of processing time, upload it to Windows Azure and then spin up a couple hundred computer instances.  The work is broken up into chunks and injected into the Azure queue and the computer instances grab them one by one, mashing on the data and then saving the results back to some local endpoint through the service bus.  Oh yes.  Then, when the batch process is done (the queue empties out), the instances are not only shut down but deprovisioned.  The solution basically means you can throw X amount of cheap computing resources for Y amount of short time and then they are gone – and you only paid for what you used, so the more you spin up, the faster it goes, the cheaper it gets.  The possibilities here are fantastic to save bazillions of dollars in storage and computing costs for things that require compute intensive processing.

These are my thoughts.  More to follow as a service offering materializes.  But I’m excited.  If you haven’t yet started playing with Azure or it’s poorly named S3 cousin, get to it!