consultaglobal

“In a new report, “clearing the air on cloud computing,” McKinsey deflates some of the claims made about the cloud, saying the buzz is disconnected from reality and the promised cost savings just aren’t there yet.”

Read Eric Krangel’s “McKinsey: Cloud Computing Still Too Expensive,” on Silicon Ally Insider.

I just typed “Cloud Slam 2009“  to do quick search and, interestingly enough, this blog happens to show up among Google’s top 5 results.

Getting back to what really matters, most speakers at Cloud Slam 2009 highlight the fact that business model innovation is as important as the underlying technology.

Here is a quick recap on what I am hearing on the intended benefits:

• utility model: no up-front commitment, pay as you go, SLA driven
• lease vs. buy: moving capex to opex; fixed to variable cost management
• speed: faster (x3-5+?) application rollout, plus on demand and simultaneous access coupled with hassle free up/down scalability
• leveraged: shared assets and federated resources (multi-tenancy mode) application reuse and customization
• focus on creating value for your business

The main service models cover three different levels of abstraction:

• IaaS, Infrastructure as a Service: you get a virtual machine to use your own software
• PaaS, Platform as a Service: includes middleware
• SaaS, Software as a Service: end users access the application

The above picture is featured by McKinsey’s “Clearing the Air on Cloud Computing,”  You can click on the picture to display the pdf file. Page two states that:

• the nascent cloud computing industry has great potential
• the industry’s “irrational exuberance” generates  unrealistic expectations.

One controversial take away from that report is that “current cloud computing offerings are not cost-effective compared to large enterprise data centers.”

At the end of the day, this business is actually ruled by SLA, service level agreements. SLAs spell out what you would be entitled to in terms of: data traffic, storage, throughput (i.e. requests per sec), scalability (instances), access to resources, security, reliability, support, reporting, etc. Additionally, corporate IT managers are wondering what to do about:

• existing legacy and packaged applications the business has come to depend on;
• creating an application in the cloud just once and scaling everywhere, i.e .multi-cloud-vendor considerations.

Quite a few of the talks deal with the notion of hybrid clouds: a combination of public and private clouds, on and off premise infrastructure (note that private clouds can be enabled off premise) which would deliver a “best of breed” model.

• The thinking is that there are private clouds that are well suited for mission critical SLAs involving high security and compliance requirements.
• A hybrid approach means being able to move applications and policies to an external cloud and retrenching as needed (mobility between clouds) to accommodate demand fluctuations.
• Hybrid clouds can also take advantage of clouds provided by different vendors; this can be set up on the basis of what the different service providers excel at.

Yesterday, Amazon made a point about their cloud being better suited than some of the existing corporate IT infrastructure. In any case and whatever the model, the top level requirements are:

• the end users’ experience is paramount and the underlying infrastructure should be transparent to them;
• the improvement in the cost/benefit ratio should be material enough to make the switch.

J. de Francisco blogging from Chicago on April 23

consultaglobal.org  :::  disclaimer & disclosure

Other posts of interest:

• Cloud Slam 2009 (1); (2); (3) ; (4)
• eComm 2009 (1); (2); (3)
• 2.0 companies leverage 2.0 technologies
• open cloud manifesto