Ben Reid discusses Service Oriented Architecture [SOA] and Cloud Computing

» Obstacles to Cloud Computing discussed

» New Zealand Government launches igovt : trusted secure online services

» Views of the Cloud (2): A topology of submarine cables

» Views of the Cloud (1): Shipping container datacenter Lego

Recently I read the sci-fi novel Thirteen by Richard K. Morgan which predicts that all manufacturing would be carried out on massive "raft" statelets which float offshore just outside international boundaries.

» Shane Robison, HP CTO, talks about "A Bright Future in the Cloud"

» Steve Ballmer on Cloud Computing

» "SaaS is not the same thing as Cloud Computing" - Response

I picked up a comment last week that "SaaS is not the same thing as Cloud Computing" and thought I'd briefly take a look at this relationship.

Generally, I've noticed that there are at least three schools of thought on the words "Cloud Computing":

• Infrastructure oriented view: "just another name for utility computing"
• Software oriented view: "just another name for application service provision"
• Service oriented View "Everything as a service (XaaS) from now on"

I guess up until now it has really depended upon what angle an individual is coming from - traditionally infrastructure folks will always focus on the infrastructure while those with a software background operate on the application level, and never the twain shall meet. However, these two views are now combined under the service-oriented view: effectively "infrastructure as a service" and "software as a service" are two deliverables of Cloud Computing, as well as "everything else in between as a service".

Therefore my opinion would be that the terms "cloud computing", "utility computing", SaaS, Software + Services and all the other current buzzwords are different views of the same underlying phenomenon: the continual drive towards ubiquitous service orientation. Choose the one that suits to hang your business proposition upon.

Underlying this discussion is the difficulty of unbundling service value and "monetizing" the investments required for Cloud Computing. Software as a service (SaaS) requires Infrastructure as a Service (IaaS) to function. Conversely, IaaS requires SaaS to deliver end-value. (No-one's going to buy infrastructure without software to run on it!). No service-oriented business model would work without both components, and everything else in between.

Generally, when discussing cloud-based monetization models I think this wider concept is helpful: If "A" aaS requires "B" aaS and vice versa, then these services must be bundled together in order to make for a business. What you end up with is a symbiotic ecosystem of interdependent services bundled together to deliver end value.

Just one example from further up the food chain: "Search as a Service" is currently always bundled with "Targeted Advertising as a Service".

» WS-* vs. REST - differences explained

I found this discussion from Sanjiva Weerawarana in the Software Development Times which discusses in plain language the differences between WS-* and REST for web services.

Personally I've got to say I fall down on the WS-* side: once you try to do anything even slightly complex (security, reliable messaging, distributed transactions, even just setting policy) with REST web services then you're going to need to recreate the functionality which the WS-* standards specify - as Sanjiva points out you're just going to end up with REST-*. It seems, er, overenthusiastic  to re-invent these standards again for the REST platform.

I think to a large degree it boils down to the business model: if it's "give away the web services for free and make money elsewhere" then REST has a place. If, however, your web services need to be secure, licenced, billable and subject to QoS obligations then WS-* is the platform to use.

» The Cloud of Economic Services

So... a blog about Cloud Computing.

First off, let's take a definition from Wikipedia (as of today anyway - these things change pretty quickly):

"Cloud Computing is a computing paradigm shift in which computing is moved away from personal computers or an individual application server to a "cloud" of computers. Users of the cloud only need to be concerned with the computing service being asked for, because the underlying details of how it is achieved are hidden."

Now, as an architect it's a pretty standard technique to look at technological entities from a number of different views. For example, one (implementation) view of the Cloud is the that of massive mirrored data centers and huge undersea fiberoptic cables, brilliantly described by George Gilder's Wired magazine article The Information Factories. Another view is to look at the Cloud* from a functional or use case perspective: exactly what functionality do the services within the Cloud provide? (Online storage, payments, salesforce automation... - basically all the WSDL out there).

However, more recently I've been thinking about a logical view of the Cloud: how do we start putting a logical conceptual structure and language around an innately nebulous concept like "the Cloud of services"?

Now, the word "Service" is a many-overloaded term , and in most conversations about "services" related to the Cloud people are talking about "web services" and "service oriented architecture" (SOA) - however, I prefer to abstract back out again to the wider economic definition of a service (again, quoting Wikipedia):

"A service provision is an economic activity that does not result in ownership ...that creates benefits by facilitating either a change in customers, a change in their physical possessions, or a change in their intangible assets."

Or, more succinctly: an economic service delivers an outcome, (which is hopefully measurable) to facilitate change. And, an economic service usually has a price of one sort or another.

My logical view would define the Cloud thus:

"The Cloud" = "the universe of all economic services"

(More than a nod to 20th century set theory here). ;-)

Here is a picture of "the Cloud of Economic Services", with a few examples pencilled in, including their economic pricing models in parentheses for discussion another time.

Now, consider the following statement, where X is any good or service:

"buy X" = "buy a licence to execute the business process "procure X"

If you agree with this formulation, then it follows that every economic good or service can be decoupled from its pricing model, and can instead be expressed purely as a licensable business process, with a single entry point exposed as a web service. That is: every economic service can be expressed as a web service:

The Cloud of all economic services = The Cloud of all web services.

Obviously the reasoning above is purely abstract, and there will be real-world challenges to expressing every economic service in WSDL (latency is the big one which springs to mind), but nonetheless I think that this forms the underpinnings of a model of how to rapidly make SOA commercially viable on a global scale - in particular, how to make a common marketplace of all (web) services in the Cloud work. More on this subject soon...

[*As an aside, being a survivor from Dotcom 1.0 circa 1995 I was disappointed when the various linguistic authorities around the world institutionalized "internet" with a lower case "i" - imho there is only one Internet (yes, there may be many little internets all operating in splendid isolation from each other, but there is only one Internet that we all connect to...), so like the Earth, the Sun and the Moon it's a proper noun! Anyway, in the same vein, rolling the argument forward to 2008: conceptually there can be only one Cloud, so I shall be capitalizing forthwith! ;-) ]