In discussions about cloud computing and in comments readers leave on my blog posts, I commonly get statements along the lines of "Yeah, this cloud computing stuff sounds great, but at the end of the day, you have to have an IT guy solving problems like they've always done." In personal interactions, I often hear this sentiment portrayed as, "Public cloud computing is fine for the SMB and startup market, but enterprises aren't ready to move to that model." The tone of much of this feedback is that anyone who advocates cloud computing is at best naive or at worst incapable of understanding the real details of IT.
So it was bracing, to say the least, to read two blog posts by executives at EMC--surely the most "enterprisey" of technology vendors extant--. Their posts present a stark perspective that cloud computing is a watershed dividing the old from the new in the way IT operates. One post is by Chuck Hollis, vice president of global marketing CTO at EMC. I always look forward to Chuck's posts--they're thought-provoking and reflect real-world feedback from his daily interaction with EMC's customers. In his post linked here, he discusses a post by Jon Peirce, vice president of Global Infrastructure for EMC IT, which presents his thoughts on what cloud computing means for the future of IT. I was not familiar with Jon prior to reading his post, but feel it is incumbent upon any senior IT executive to read and assimilate his message.
Essentially, Jon makes the following points:
* IT has traditionally been given a fixed budget and decides which of the many business unit demands it will satisfy with that budget. Because it plays a rationing role, IT seeks to reduce demand for IT resources. Said another way, unlike businesses, which see demand as an opportunity to expand market share, IT has traditionally seen demand as exacerbating its rationing challenge and therefore something to be resisted. Thus, IT's reputation as "The Department of No" emerges. Rationing also means that business units are dissatisfied with IT service levels and have a strong motivation to find other ways to satisfy their IT requirements. In the past, it was extremely difficult to find those other ways, but with the rise of public cloud service providers, there is now a convenient way to bypass IT.
* IT needs to change its self-image from a project manager mindset to a service provider mindset. It needs to welcome business unit demand and view itself as a service provider, ready to deliver resources to any and all that desire computing power. For sure, it needs to step away from the role of arbiter among resource demands, because that's a losing game now that IT is no longer a monopoly supplier.
* IT needs to adopt chargeback based on granular services. Business units need to have accurate signals about how to evaluate their use of services versus the benefits derived by deploying those services. Cost is the signaling mechanism used in businesses, so IT has to be prepared to offer chargeback. Jon put it this way: "Business leadership needs to be held accountable for IT service consumption." In some very mature organizations, this may be achievable with "show-back" of costs to the consuming business units, but in most enterprises, "charge-back" will be a necessity."
Jon's larger point is that IT needs to transform its culture from an insular, technology-focused monopoly provider to that of a market-serving, demand-generating, business enabler based on delivering IT service capabilities.
There are a couple of assumptions in Jon's recommendations that I believe are quite challenging for the typical enterprise IT organization, and I wonder how EMC has addressed them.
First, as Jon points out, for this transformation to occur, "CFOs will need to embrace a new funding model which places the financial responsibility for consumption on the shoulders of the business consumers."
A different way to say this is that service provisioning requires building capability prior to demand arrival, and CFOs must be ready to fund capacity to enable service provisioning. Once the capacity is in place, IT can get on the task of selling services to business units. This model is, of course, well-established in manufacturing--you have to have a factory in place to build the widgets before you can sell them--but one may expect that the average enterprise IT shop is going to have a lot of work ahead of it to sell a CFO on financing an information factory.
Second, Jon feels that internal IT can deliver cloud computing at a lower price point than a public provider. He writes, "IT shops can introduce multi-tenant, automated, highly secure, tiered, self-provisioned capabilities that can deliver lower unit costs and better SLAs than what's available in the Public Cloud for IaaS [infrastructure as a service] and PaaS [platform as a service]."
I'm not sure I understand how internal IT groups can achieve lower costs than public providers. I can't think of a single cloud computing cost input that, in practice, is less expensive for internal IT organizations than for external providers, and I'm not convinced that the profit margin that public providers seek will tip the scales toward internal IT, except in rare cases.
When I discuss this cost advantage topic with private cloud advocates, ordinarily the conversation starts with putative internal cost advantages, and, when I pose the question of relative cost advantages, the discussion quickly shifts toward compliance, security, privacy and (as in Jon's statement) SLAs.
Frankly, in my view, this is where the battle of private versus public clouds should be fought: on what non-cost requirements certain applications have, and where the best place to address those requirements is. Trying to assert that one's internal IT can deliver computing less expensively than, say, Amazon Web Services, is likely to lead to business unit disillusionment with the internal offering, once the actual costs are sorted out.
Far better will be for IT to outline services in which it can offer undeniable advantage, like certainty of compliance, privacy or business domain knowledge. The corollary to that is, IT should recognize application profiles that do not require those elements and seek to place those workloads in the most cost-effective location. Absolutely crucial to this application placement decision process will be an automated deployment and governance system that can in real-time assess application characteristics and dynamically choose where to host it.
It's often been said that only a vociferous anti-Communist like Richard Nixon could have created the US rapprochement with the People's Republic of China because only someone with Nixon's background could convince people of a similar mindset that the initiative made sense. Not to draw too close a parallel, but enterprise IT voices like Chuck's and Jon's are powerful persuaders to mainstream IT shops. Only when someone with impeccable credentials points out that the established methods are no longer workable will many begin to evaluate new options with an open mind. If you are a senior IT executive--or a senior finance executive--you owe it to yourself to see what your peers say is the future of how computing will be delivered.
Bernard Golden is CEO of consulting firm HyperStratus, which specializes in virtualization, cloud computing and related issues. He is also the author of "Virtualization for Dummies," the best-selling book on virtualization to date.