Talk:Cloud computing

Concerned about scope and precision

To me, cloud computing is a form of distributed processing that emphasizes the computing resources, not the applications, beyond the technologies (e.g., intelligent DNS redirectors) that send transactions/sessions to resources. There is absolutely no non-buzzword reason to link SaaS, for example, with clouds. Clouds, grids, and other distributed infrastructures that are ad hoc or demand-driven are radically different than SaaS built for high availability. So far, when I've done SaaS architectures with any concept of a service level agreement, as in healthcare, the computing infrastruture is far more specific than a cloud, with extensive load distribution, fault tolerance, and capacity planning.

I'd hate to see us drifting into one of the "Web 2.0" styles that makes everything so general that it provides no engineering guidance. Howard C. Berkowitz 19:11, 11 March 2009 (UTC)

Everyone has a different idea of what cloud computing is - the article developed at Wikipedia is somewhat a consensus that covers/satisfies most views. Your "evolved grid" view of cloud is quite limited compared to others and the average reader has been conditioned to think Google Apps and Salesforce when they hear the term. How technical is our audience really? Sam Johnston 19:38, 11 March 2009 (UTC)

Speaking as a Computers Workgroup Editor, CZ articles will be technically correct. We are expert consensus based, not general consensus based. That isn't to say that things cannot be written to be understandable by people with limited detail knowledge: our general target reader is a college undergraduate. One of our goals is "deconditioning" marketingspeak into accurate content. Our definitions are written (or approved) by subject matter experts, not "average readers".

Last year, we went through some of this with an Eduzendium (educator-guided student writing assignments), and there was a fair bit of insistence on rigor when dealing even with evolving technologies such as mashup, Ajax, etc.

Twitter and blogs are not preferred sources, and, when dealing with the industry press, it has to follow CZ: Neutrality Policy.

To convert this from WP to CZ style, the first steps are to give even conflicting definitions, and then compare and contrast. Real-world issues like availability/service level agreement must be covered.

Salesforce is SaaS, but with some fairly critical service delivery requirements. First, explain how clouds work without going off into the application level. Yes, failover and load distribution are things to be addressed early in the article, not SaaS that could be delivered with a distributed computing system totally under the control of a single vendor.

Let's start with Infrastructure as a Service and move from there. Howard C. Berkowitz 20:15, 11 March 2009 (UTC)

Ok I've got a train to catch in <24 hours so don't really have time for this right now. I was planning to iteratively improve the Wikipedia content without Wikipedia interference over time (was there really a twitter ref in there?) Sam Johnston 20:44, 11 March 2009 (UTC)

Supercomputing

I've moved "Cloud computing aims to apply the power of supercomputers—measured in the tens of trillions of computations per second—to problems like analyzing risk in financial portfolios, delivering personalized medical information, even powering immersive computer games, in a way that users can tap through the Web."

out of the article. Transaction processing appears to be one of the more common services; examples given such as SalesForce and Google Apps hardly need supercomputer computational power, but highly parallel client/server processing. Some cloud applications may be computationally intensive, especially using grids, but the reality is that not that many applications exist to make use of the traditional computationally oriented definition of vector processors or massively parallel processing of specific applications. Having high volumes of individually compute-moderate transactions, however, is very reasonable.

Storage distribution needs to be discussed in more context, including the federated versus distributed model, replication and reliability, and synchronization.

I've also put references into CZ standard form. Howard C. Berkowitz 20:52, 11 March 2009 (UTC)

Product examples

We'd usually like to have the significance of commercial products explained, perhaps with reference to neutral sources, to avoid the appearance of advertising. I removed the text below; it certainly can go back with more context.

"In some services, such as Nirvanix, the system may span multiple data centers or even continents.

Examples

• IBM’s Blue cloud
• Amazon Elastic Compute Cloud
• Microsoft Live Mesh
• Apple Mobile Me

Howard C. Berkowitz 21:05, 11 March 2009 (UTC)

All of this content predates me - we can work through it later or discard it and start from scratch. Sam Johnston 21:11, 11 March 2009 (UTC)