The concern and anger unleashed by this weeks leaking (or maybe the writing) of the 'Cloud Manifesto' has managed to drag a blog post out of me. James Urquhat's take (Cloud Manifestogate) seems pretty on the nail too me - the document is really a collection of high level principles (unless there is much more which isn't public), and far from a technical standard. The moves do unfortunately look rather a lot like the fun and games which revolved around SOA standards between OASIS and W3C at key times. The seeming chilling effect on activities like the CCIF could certainly be negative - I guess we'll see this week.
Taking a step back from politics however, it's a little hard to see where the main impact of such alliances (and potential standards might be). 'Cloud computing' means many things to many people, but at the risk of putting some noses out joint, most of the discussion right now is around infrastructure on demand clouds - CPU cycles, Storage, Bandwidth, OS variants, CDN capacity etc. then followed (but at some distance) by Data and Application Clouds. While I fully believe all three levels will see radical changes - many of them due to the emergence of a whole host of standards and common formats in various places - I doubt de jure standards will have much impact at least in the medium term.
First up, at the infrastructure level, clouds have already meant a radical shift in the convenience of using hosted services - Machine AMI (Amazon) and automated build scripts from all sorts of packages (e.g. Rightscale) make a huge difference to how easy it is to port services from one host to another. Standardizing APIs to access compute resources saves time and resources - great when it happens - however the number of these APIs seems unlikely to explode exponentially (leave that to data and apps). Given the similar hard/software profiles emerging it seems unlikely the APIs are going to be a major barrier to cloud-to-cloud mobility.
There is a great thread on cloud computing on Fred Wilson's Blog (Can Amazon run the Table...?) which makes a lot of points on the Pros and Cons of various clouds and the strategies of the players involved. From the infrastructure cloud players id's expect:
- Them all the shoot for scale.
- Them all the shoot for geographical presence (big apps want seamless movement of points of present around the globe).
- Them all to shoot to support most standard hardware/software configurations.
Seems to me that players yet to emerge will come from existing hosting providers which might be able to use cloud software to turn themselves into alliances of clouds - all of which support identical services - and at the same time provide excellent global coverage. Companies like 3tera and Rightscale are in my opinion also dead on the money - given enough developer time you can work around most differences in the majority of infrastructure APIs and abstract them for your customers.
For application and data clouds things are much more complex - I'd expect to see an amazing amount of standardization in a lot of places as lead players open up APIs, Services and hosted solutions which set the standard pretty much everybody will start to follow in that area. It seems unlikely this will be driven by a standards body but evolve differently in each area (but quickly). Lets how patents, copyright and other disputes don't get in the way of data interoperability too much!
-
Comments