Overall McKinsey’s “Clearing the air on Cloud Computing ” is a good effort – that addresses definition, economics, pitfalls etc (see here for alternate copy). I happened to chat with a McKinsey consultant many months ago to provide input to this report. Being a practitioner in this space, I don’t agree with some of it, and in this post will give you my opinions (yes, fact-based but nevertheless opinions) that can clarify and hopefully be helpful to a few.

To cut to the chase, the article makes 3 key points in my opinion:
1. Does inject a dose of practical realism for corporate CIOs who are considering the public cloud (in McKinsey’s definition private clouds are simply extended virtualization)
o Cautions CIOs of large corporations that public clouds may not be Nirvana, though it does not make the point that corporate IT use is not a “one size fits all”. I disagree with the analysis in that economics are use-case dependent and enterprises can certainly benefit from public compute and storage clouds depending on the application.
2. Recognises the value of the public cloud to SMB/SMEs. 100% agreement.
3. In a roundabout way the article acknowledges that “private clouds”, which are not clouds but only “aggressive virtualization on the traditional enterprise IT stack”, can deliver significant practical benefits to the enterprise. Relatively, more benefits than the public cloud. (I suppose ParaScale, publicly announcing a product after the report was written, and others which fully leverage a commodity (vs. traditional) IT stack – can provide even better benefits? :-))

But don’t take this study as the last word (sorry, Big McK). While the cloud is defined as “ hardware-based services offering compute, network and storage capacity” … the economic analysis is only based around EC2 and public cloud computing. This is flawed for 2 reasons.
o There are compute applications that can be highly seasonal or temporary that will have unbeatable economics in a public compute cloud. This report does not take this view and makes “big brush” assertions around better infrastructure utilization.
o Public storage cloud economics (use-case dependent again)are different from public compute cloud economics. This report is only assesses compute clouds, with a passing mention about storage clouds.
o (Cloud networking is too nascent and I am still trying to understand how different it is from virtualization-aware 10gE)
Other points in the article
o Calls out “virtualizing servers” as a very practical alternative to cloud computing – well both internal and external cloud computing depends significantly on underlying virtualization technologies. End-users will benefit from understand this technology underpinning versus viewing cloud computing as completely unrelated.
o “Publicly-announced private cloud computing environments are not by definition true clouds because enterprises can’t avoid capital expenditures”. This is how McKinsey has subjectively defined it. ParaScale enables both public and private clouds – so we don’t have an axe to grind. In our view, companies like GE (who has announced their private compute cloud) easily has scale to build an internal cloud and have all the various BUs and user-groups benefit and just pay-for-use. Seems like a cloud to me.
(see here for ParaScale’s cloud definition)
o Most of the analysis bashes Amazon EC2 and focuses only on the public compute clouds(since internal clouds are not clouds) and ignores storage clouds.

Ultimately, a better articulation from 3rd parties can only help, especially coming at it from a business angle versus vendor or industry analyst perspectives. And being that this is from McKinsey, C-level execs will get exposure to this topic … and as the adage goes “there is no such thing as bad publicity” in terms of getting the word out on cloud computing :-). And perhaps McKinsey, Bain, BCG, Booz Allen (full disclosure,I was a partner at Booz Allen) or some other business consultant will have an even better business/analytical view in the future.

Sajai, CEO ParaScale

PS: After this blog was posted, VMware just announced “the industry’s first operating system for building the internal cloud—VMware vSphere 4″. Interesting that McKinsey came down on one side of this religious debate around public/external vs private/internal clouds saying there is no such thing as an internal or private cloud, that it is only virtualization. And now VMWare the leader in virtualization, announces software to build an internal cloud.