Yesterday I was hanging out at the Microsoft office in Los Angeles for an “Architecture Council” meeting on the new Cloud Computing platform. You can find the slide materials on the Strategic Architecture Council blog.
Overall, a worthwhile session that you should attend if you get the chance in your city. My favorite local evangelists were there, including presentations from Woody Pewitt, Kevin Boyle and the delightful David Chou.
We discussed what Microsoft considers cloud computing, the trade-offs between “on-premises” and cloud apps (e.g. on-premises apps have more control, customizations vs. cloud apps that are cheaper but sacrifice control and customization). Also liked the point that we shouldn’t see cloud apps as simply current in-house apps that we shove to the cloud (ala people throwing SOAP interfaces on existing APIs and claiming to be “service oriented”). We should be identifying a new class of apps that exploit the cloud in a safe, efficient manner.
Woody presented Live Mesh which I’m aware of, but haven’t spent time actually playing with it. As with many demos for these new MS tools, the focus was a bit more consumer-oriented. I’m still trying to get my head around any use cases for a tool like this in enterprise software.
David presented on .NET Services (formerly BizTalk Services) and once again inspired me to install the latest bits and walk through some scenarios. Damn him for distracting me from other things. Finally, we covered Azure and the types of web based apps we can now host in the cloud.
David discussed a few use cases for cloud services that got me thinking about how I’d use this in enterprise scenarios. For instance, it makes a lot of sense to me to expose my organizational’s reference data (product list, sites, etc) as cloud-based services that folks can use in their own apps that collect data points about us. Why should they maintain their own tables with copies of our reference data if we can provide it in a public internet service? By putting this (data+service) in the cloud, I have one less hole to punch in our external facing infrastructure.
Clearly lots of this stuff will change prior to the formal release, so for me, these technologies fall into the bucket of “have enough knowledge to be dangerous” and no more. I need to be able to talk about these technologies and hack up small demonstrations, but most importantly, I want to be able to know when I should consider these technologies in future projects for my company