Wednesday, April 7, 2010

Clouds, Stages, and IT Spending

There is a lot of talk about what is or is not “cloud computing” and whether such a thing as a “private cloud” does or does not exist. In the end I believe there are many forms of cloud. The question examined in this post is “what impact will a move to cloud computing, in stages, have on IT spending?”

Recently Michael Neubarth wrote another fine article. Its title is “Are Private Clouds Hogwash?” For some reason for a number of people the very idea of a “private cloud” is disturbing and disorienting. Neubarth quotes a number of the literati on the subject. Most say cloud computing must be outside the enterprise or clouds require a scale of computing and mutually-balanced workloads that a single enterprise cannot offer. Maybe, though this argument seems to fly in the face of the origins of many cloud providers (Amazon,, IBM).

The private cloud naysayers do cite a concern that I share. Some people are simply calling their traditional data centers “private clouds” and declaring victory. But, simply employing virtualization, web-based computing and other technologies doesn’t make a cloud. Neubarth quotes P Laudenslager on his blog, “Calling [the use of modern technologies alone] a private cloud is like drinking alone and calling it a private party." You can’t just put lipstick on a data center. There is a required aspect of cloud computing that is by-the-drink, load balanced, on demand, scalable with the financial, service management and administrative practices needed for businesses to rely on such capabilities.

Still, one might well be confused reading the complaints about the term “private cloud” at all versus seeing where major players are placing their bets. IBM, Acadia (the joint effort of EMC, Intel and Cisco), Google, Amazon, Microsoft and others might disagree. Acadia’s stated purpose is to offer private clouds as evolutionary pathways for IT. Many readers of Rock Icons and IT Thought Leadership feel as Judy Collins did when she wrote:

I've looked at clouds from both sides now
From up and down and still somehow
It's cloud's illusions I recall
I really don't know clouds at all

The research firm Saugatuck Technology has a take on Das Kloud that is fairly centered. In their research alert Cloud IT: Stages of Simultaneous, Disruptive Growth and Change” they describe cloud computing as “The Cloud. IT as a Service. Platform-as-a-Service. Software-as-a-Service. Hosted services. Managed Business Services. These and many other capabilities and offering types comprise what we have labeled “Cloud IT.”
It’s a deliberately inclusive definition. They go on to say that “all aspects of user IT and business operations are rapidly melding into an inextricably interwoven, and inter-dependent, series of capabilities that are delivered and used as one or more services via one or more networks. [Saugatuck] no longer sees easily-definable boundaries between IT and business; just as [they] will soon see the disappearance of boundaries between traditional (on-premise) and Cloud-based business and IT.”

Saugatuck’s point is that user IT and business operations are rapidly becoming intertwined “into an inextricably interwoven, and inter-dependent, series of capabilities that are delivered and used as one or more services via one or more networks.” It’s a practicalist view, not a purist view. They see that the business/IT boundary is blurring as well as traditional, on-premise and off-premise IT. By extension clouds must reach from outside the business into the business.

Saugatuck also sees the use of Das Kloud evolving in four stages within enterprises. The first is the Point and Supplemental Stage where the majority of Cloud IT exists today for users and providers. In this stage cloud infrastructure offerings and use is “filling in gaps, augmenting existing system and process capabilities.”

The second stage is the Developmental Stage. Cloud computing makes a practically limitless scope of software development resources available to users and vendors alike, and very inexpensively. Software developers no longer have to rely on dedicated internal resources to code, test and refine. Resources are available whenever (and wherever) needed, and can be turned on and off as required.

The Internal Growth Stage is the third stage and involves significant development, deployment, and use of Cloud IT within user organizations as part of overall IT strategy and operations and includes private and/or internal clouds.

Saugatuck’s vision has a fourth stage, Hybrid/Interwoven,” wherein customized combinations of traditional, Internal/Private Cloud, and third-party Cloud IT are delivered and used. Saugatuck ties it all together with this model:

--> -->
Working on Step 2

If we take Saugatuck’s estimates and stage descriptions as an example what are the implications for IT spending? Recall the basic concepts behind the IT funding model:

  1. Systems built this year must be run and supported over time
  2. Left unchecked, the compounding effect of increased operations and support costs inexorably crowd out further new systems spending
  3. Because businesses cannot tolerate static systems, because system change and new challenges emerge, constant efficiencies are needed in operations and support costs to make budgetary room for new systems development

As described in Holding the Line on IT $pending, many companies are pursuing Saugatuck’s first stage of Supplemental Point cloud solutions in their campaign to hold down operations and support costs. These often include solutions as storage and processing. In the case of SaaS, many customers have deployed focused “point” solutions focused on key business. Saugatuck says these “solutions differ in concept very little from any early-stage, newly-introduced IT, with the possible exception that they are exceptionally low in cost to acquire.” And that is their point, budget reduction in operations to enable development spending.

The second stage deals with cloud computing and systems development. In an example of Platform as a Service (PaaS) multiple providers are coming to market with entire offerings for systems development in the Das Kloud. IBM and Microsoft are both investing heavily in Cloud-based development platforms environments and services for customers and their ISV and development partners. The Cloud also makes a practically limitless scope of software development resources available to users and vendors alike, and very inexpensively. This pattern will reduce the cost of developing systems and thus may ease the demand to reduce unit costs in operations and support, in the short term.

However, as cloud-based development becomes standard all competitors will have the same capabilities and the old pressures will resume. IF, as some claim, cloud-based development is a productivity enhancer for systems development it will bring systems to production faster. This will increase the pressure on support and operations to reduce costs as their workload will increase without resources to handle it. This will push those organizations farther into the cloud-for-cost-operational-reduction camp.

Saugatuck’s third stage, the Internal Growth Stage, is the significant development, deployment, and use of Cloud IT within user organizations, as part of their overall IT strategy and operations. It is a reaction of the success of the first two. If the Stage I, or Initiation Stage uses of Das Kloud reduces operations costs and eases development then everyone will want to be on those bandwagons. This typically results in a rapid increase or “Contagion” in use of computing and, in this case, a rapid increase in cloud-related spending.

Nolan's 4 Stages of EDP (IT) Growth


I’m not particularly concerned with whether or not there IS such a thing as a “private cloud.” I think the description is apt enough. It describes how many very large organizations today, and smaller ones in the not-too distant future, will use cloud technologies. I’ll leave the nomenclature battle to others.

I do believe there is a driving requirement to reduce IT support, maintenance and operations unit costs day in and day out, week in and week out, year in and year out. The Cloud appears to offer assistance in making reductions without sacrificing delivery and quality. As and if that proves out expect your business to use much more of it in the near term regardless of the name you put on it.



  1. Russ Bostick writes: The best exposition on the topic I’ve read. Only one objection. I suggest that you change the highlighted words below.

    Because businesses cannot tolerate static systems, because system change and new challenges emerge, constant efficiencies are needed in operations and support costs to make budgetary room for new systems development.

    A better phrase would be “Because business models change”... No business wants IT to change it’s systems just because they are ‘static’. Change development to “delivery”. Cloud systems are acquired and integrated, not developed.

    I would have posted this on the comments, but the Luddite CIO here blocks a lot of social media. :}

  2. Guiseppe Uslenghi writes:

    This cloud concept seems to me quite similar, in principle, to the IT outsourcing concept pioneered by Perot / EDS a few decades ago: centralise infrastructure and resources for better economies of scale, and cost efficiency.

    Of course available technologies today are quite different form then, allowing for a much more articulated approach....but maybe there are some lessons to learn from the EDS story (and collapse) that may be applied here.

  3. One supposed difference between today's cloud and yesterday's outsourcing is that today's cloud should be offering more opportunity for control (and responsiveness) despite the fact that today's cloud resources aren't owned by the user. Yesterday's outsourcing model tended to require more accommodation to the vendor's own strategy.

  4. Judy Collins may have sung the song, but I believe it was Joni Mitchell who wrote it!

  5. Y'see.... I really DON'T know Clouds, at all, sorta'....

  6. The essence of cloud computing is that it is a commercial offering. Everything else is just old hat, which is why some people are enthusiasts and some annoyed. Software as a service implies at least that there is a service – and therefore is something which customers buy, and which can be described coherently from the customer’s point of view. How the service is assembled may involve a bunch of old stuff (configured technology widgets) and of course whether this is “private” or not (and what would it be if it wasn’t private – commonly owned??) is irrelevant. What is important however is that the supplier of the cloud services has to have a coherent, workable and sustainable commercial model. Which means not only being able to manage the technology capacity, but also all the administrative and commercial processes required to sell a service. The reason old EDS wasn’t doing this before was twofold. Firstly, whoever owned the kit, the kit was typically customer specific; the customer needs more capacity, the supplier buys more capacity for that customer and installs it in the corner of the data centre that will forever be that customer’s corner. Secondly, many outsourcing contracts in effect had a guarantee built in, that in the event that costs were more than were expected (with suitable caveats) they could go back to the customer and claim back the overrun. This isn’t buying an end to end service, it’s buying bits and pieces (some of which might be sub services) and cobbling them together. The guy who sold you a pair of shoes doesn’t come back at the end of the year and ask you for another $20 because the cost of his air-con went up, but that in effect is what EDS would have done.

    In order to sell something as an end to end service you need to take the commercial risk. You take the design risk, you manage the overall capacity, you offer the SLA’s associated with upgrading storage or processors, and you make the profit. You don’t just offer to store someone else’s kit, and run it for them. I think that what really distinguishes cloud computing from anything else is the management of that commercial and technical risk, and this will de facto be hard to do if all you have are a few servers in a corner. It is essentially a matter of scale, not of ownership (private vs. something else) as well as the right management tools to track the costs and bundle them up (not just “pass them on”).

    Whether a company chooses to implement cloud computing internally (i.e., develop all the systems needed to sell such services internally) will at least be a question of scale (according to McKinsey Deutsche Telekom has done this). But actually there isn’t anything special about cloud computing from a make vs. buy perspective – you go through very a similar process whatever the service being procured.

    By the way, what’s this “das Kloud” stuff? Cloud in German is (die) Wolke – and cloud computing is now known as “Cloud Computing”!

  7. I think your focus on the commercially definable service is the core of it. I am reminded of Geoffrey Moore's concept of "the whole product" as described in "Crossing the Chasm." Its not just the physical technology, its how its offered, supported, administered, paid for and changes over time.

    Although much of the macro sense of Cloud Computing can be viewed as "old hat" I do think that in the 1980's we had the ability to conceive of such services. Now, building on layer upon layer of technology and "whole product" advances we can realize it. Perhaps its a small point but I don't see how in, say, 1990, today's Cloud Computing could have been provided with any economic sense.

    ps - "Das Kloud" is just playing around with "Das Boot." Do I HAVE to capitalize "Cloud Computing?"

  8. I fully agree with your last comment - I don't think the tools were available to sustain a competitive cloud service in the 1980’s or 90's, but my main point is that to argue that concentrating on the underlying technology is to miss the point. For certain, virtualisation helps enormously in being able to offer services which rely on underlying shared infrastructure, and the ability to offer increased storage (say) quickly and effectively is also a technical precondition, but what makes it different from traditional outsourcing is the ability to take risk. I also remember doing a bit of work for Ameritech in 1988 (when I was in the NNC Chicago office) whose vision was “Bandwidth on Demand”, but they didn’t really know how to deliver it, short of provisioning for maximum demand, charging for what was permissioned, giving permissioning tools to the client and hoping the economics would take care of itself. Fat chance!

    "capitals are for neurotics and germans": ee cummings