State of Change, Chapter 18: The Elastic Data Center
There is no “the cloud.” There is one Internet, and by way of the virtual connections between sites, there is one Web. There is certainly one global economy, any division of which at any scale can be declared a “market.” But insofar as the computing power, storage, and bandwidth required to run your business is concerned, “the cloud” is quite literally neither here nor there. The IT resources used to run your business are still your resources, whether you house them within the data center on your premises, or you lease them by the month or by the hour.
There is, for any enterprise, one network, within which is one data center. (Hopefully not more than one, though if so, my sincere apologies.) Viewed from one aspect, the whole discussion about “the cloud” is merely a little detail about its location. The great achievement of the Internet, it has been said, is that it makes the location of information an insignificant fact. Ironically, it’s the degree to which it is insignificant that has become the most significant fact of technology since the advent of the assembly line. Since information resources can be off-premise as well as on, and anywhere in the world is as good as next door, technology architects have endeavored to capitalize on your newfound ability to move your data center, or significant parts of it, from here to there. “The cloud” is the “there” in the argument.
“The cloud” is presented to businesses for the first time as though it were “the world,” or the whole of something that’s outside your corporate firewall.
“The cloud as a place to store information, to build applications, as the place that has the platform, is in fact very different from the Web, because the cloud is replacing the form factor of server farms,” remarks Anna Rosenman, senior product marketing manager for Salesforce Chatter. She continues:
The cloud is where all of your technology lives, whether for your employees, your customers, or your partners. And people choose how to get there...
I always think of the Web in the same way as your phone is a vessel. People use each of them differently, depending on the form factor. If you’re outside, you’re going to use your mobile phone [and] your Southwest Airlines application or your Chatter application. And at your computer, you’re going to click on the Chrome button or the Firefox button and access all this information through the Web.
More often recently, consumers perceive (with more than a little help from some vendors) “the Web” as the place they contact with their browsers, and “the cloud” as the place they contact with their smartphones. Even though many mobile apps are essentially branded skins atop mobile browser cores, consumers have already been conditioned to perceive the cloud as an ethereal, boundless, all-encompassing information ecosystem. This perception bleeds into the executive or “C-level” perception, if you will, of the cloud as a space where servers and perhaps their administrators can disappear out of sight, but not out of mind.
A CoreSite data center facility in 2014.
The Hybrid Myth
No doubt by now you’ve read that the “first question” on your mind when evaluating a cloud-based deployment either is, or should be, whether you should choose a public or a private cloud, or something in-between. Hopefully if you’ve read the many parts of this series leading to this point, you’ve already come to the realization that this is a false concern. It’s a question a salesperson would prefer you ask yourself, to make it easier for you to march yourself in the desired direction down the sales funnel, to make it easier to eventually sell you something.
The reality of today’s emerging business computing environments is defined quite differently. Today, the power that you would traditionally be generating inside your on-premise data center with your own equipment, to process data and produce documents and communicate, can be leased as a service from an outside provider. Eventually — hopefully soon — we’ll stop appending “-as-a-Service” or “-aaS” to the end of the names for leasable resources, though it’s difficult (especially for Americans) to overcome the compulsion to pin tails on things unnecessarily.
“If I’m a CMO and I decide that I really need to enhance my mobile customer engagement strategy, and IT tells me it’s going to be nine months until they can get to me, do you think I’m gonna wait?”
The technological change itself may be so rudimentary as to be inevitable; were it not for a few global recessions, the whole cloud phenomenon might have taken place in the 1990s. But as you’ve seen, this simple, emerging truth is calling into question the business models of entire industries — often in different ways, but usually for one or more of the same reasons:
Because cloud services are customer-provisioned, they become easier to manage. Reducing the complexity of IT management in turn may reduce its priority in the executive mindset. The devaluation of IT that often results calls back into question the original goals of information services integration, and the entire bit about the aligning of IT goals with those of the organization. If IT management becomes an operational expense like janitorial services are an operational expense, many businesses find themselves pondering, perhaps information strategy should become the responsibility of other capital divisions, such as marketing. A January 2012 webinar by Gartner VP Laura McLellan predicted that by 2017, companies’ CMOs will spend more on IT than the CIO. That’s not necessarily because marketing budgets are rising, but since authority in organizations typically follows the money, the declining cost of IT alone could precipitate a shift in business priorities.
Because cloud services can be shared (even the ones on-premise) with other service customers, the very definition of what constitutes “the firewall” in an organization becomes obsolete. At the onset of the Web era, some companies redrew their departmental boundaries to align with their domains and subdomains. But as the Web evolved beyond the need for subdomains, companies that did align their IT and corporate strategies began to perceive the firewall as the perimeter of more than their data center, but their organizations as well. Now that business partners can conceivably share the same physical, and in some appropriate cases, logical resources by leasing them from the same provider, the idea that a business should be modeled after a castle or fortress appears antiquated. The need for intensive, incremental, auditable systems of recording business transactions comes into question — and with it, the justification for as much as half of those business’ existing IT.
Because the cloud option raises — in some cases, by law — the question of how companies maintain privacy and accessibility controls over the data they share with cloud services, in many cases, it is compelling (and even forcing) companies to consider privacy policy seriously for the very first time. In the course of the discussion, departments and divisions are taking responsibility for their data in ways they have not done before, especially in an era when data was by definition on-premise, and by experience difficult to maintain. To the extent that personal data involves customers, the marketing department is taking charge. And to the extent that it involves employees, the human resources department is playing a more active role — in many cases, assuming leadership responsibilities for ensuring compliance and for maintaining governance policies, in organizations where more personal action is needed. As a result, data is being treated for the first time not as some residual product of everyday business transactions, but instead as a critical asset of the company. And this is making executives and boards of directors rethink the question of how funds are raised in the collection and harvesting of this data.
Because cloud services can be provisioned for limited, short periods of time, scientific applications that may only be applicable periodically, and production applications that only need to happen on occasion rather than perennially, instantly change from impossible to afford, into trivial to procure. As a result, businesses whose products may include computation itself are able to employ cloud infrastructure on temporary schedules, like factory floors that wink out of existence when they’re not utilized. This not only changes existing business models directly, but can bring new and unforeseen businesses into existence. These new businesses may themselves be competitors to established firms with older business models. Suddenly these older firms find themselves having to innovate to survive.
In this light, it becomes obvious that the first question on your mind isn’t really the one that vendors might hope it would be. There is a related question about how much responsibility for governance and control that you and your company would want to delegate to third parties, and that’s certainly important. But it does not correlate to the public/hybrid/private cloud architecture question to the extent you might think.
“For enterprises of scale, the cloud computing of the future will be a hybrid model. There’s no doubt in my mind,” says Ric Telford, IBM’s vice president of cloud services.
There will still be data centers. Large companies are not going to get away from running data centers. But data centers will be transformed over time to be a cloud delivery model, wherever it makes sense — so think of that as a private cloud. And some applications, and parts of business processes, will be sourced to public cloud service providers like IBM, and you’ll need to integrate the two and have hybrid cloud.
You met IBM’s Telford in an earlier article in this series, discussing the effect of what I call “cloud dynamics” on the structure of enterprises. There also, he referred to “enterprises of scale” as larger organizations, and distinguished them from small-to-medium businesses (SMBs) not just in terms of size but also needs. Many SMBs, he said, want out of the IT business altogether, and to at least have what I call “the nuclear option” on hand.
But as more servers adopt virtualization at a low layer, regardless of the scale of the business in which they reside, the option to use cloud resources anywhere will be made available in terms of quantity. Data center environments will already be cloud-ready, and as a result, the controls for provisioning and administering hybrid clouds will already be in place. So for IT departments, hybrid cloud will be the reality anyway. Technologically, the changes are already being made. Ric Telford:
Cloud just gives you more options. You’re not locked into one paradigm of IT any more. The early days of cloud have been around optimizing IT and reducing costs, but most of the conversations we’re having now have to do with how to leverage this more flexible delivery model for business differentiation, because it can allow companies to do things they couldn’t do before.
I’ll give you an example from the media industry: One of the computationally exhaustive, expensive tasks they do is digital rendering, for animation or graphics. Digital rendering only happens at certain times of the year. But do you really want to have 10,000 servers that you use three times a year to render a movie? Or would you rather get 50,000 servers for a week and do it in the cloud? It’s going to change a lot of business models, because they now have a lot more options and flexibility from IT than they’ve had in the past.
Another example Telford offers from the engineering industry: Many firms have historically been unable to bid on certain government contracts, because they lacked the computational resources to run the simulations they would require to design and innovate entirely new parts that meet or exceed government standards. Until recently, acquiring the infrastructure necessary to accomplish this was a capital nightmare. Now it’s just barely a task.
Amazon Web Services (AWS) is arguably the pioneer in offering “burst mode” resources. It’s also the symbol of what is typically called “the public cloud,” although again, there’s really no such single thing, like “the public treasury.” AWS is available to the public, which is the reason for the distinction. A private cloud, by contrast, is made up of a data center’s internal resources that can be provisioned for tasks and applications in much the same way as an infrastructure service (IaaS) — provided, of course, the IT staff doing the provisioning is on premise. But cloud infrastructure platforms such as Cloud Foundry and OpenStack are just as essential for administering a firm’s services when its resources are entirely on premise, as they are for administering distributed compute and storage resources that are mainly leased from public services like AWS.
It’s through the type of strategic architecture typically called “hybrid cloud” that burst-mode applications become possible, such as the ones Telford described. Amazon’s principal data scientist, Matt Wood, offers us another example from the pharmaceutical realm, where clinical researchers are only granted the funds to conduct experiments only after it can be demonstrated that the risk level is relatively low:
The risk of failure of that experiment in a capital investment, statically provisioned infrastructure is very, very high. You’re making that commitment to run those servers, and keep that storage online, for three to five years, whether the experiment works or not. On AWS, in a utility environment, if that experiment doesn’t work out, you just turn off your infrastructure and you stop paying for it. Or you evolve your infrastructure; you put in a different class of storage, a different type of CPU, higher memory. You have a lot more freedom to experiment as you continue to iterate, and we see that as very, very enabling across vertical markets for customers.
We’re also seeing... companies wanting to be able to take the data that’s being generated by their applications and their customers — that business line-of-sight data — and draw out actionable information. Obviously, that’s compute- and storage-intensive as well, and has a high cost of experimentation. AWS completely removes the cost of that experimentation; you can iterate on that data more quickly, and start to move more quickly to enable your business by using that data, rather than leaving the value of that data on the table if you don’t have the resources to start to unlock it.
The C-something-O
The CIO role was created in the late 1970s as a suggestion by Dr. William Gruber and William R. Synnott. They envisioned a strategic leader, as opposed to a manager, who could take charge of information as a key strategic asset for her organization. The widespread success of the CIO concept had an unforeseen side effect: the notion that a person at the executive level should be “focused” (there’s another word you hear all too often in marketing) on one aspect of the company, to whom the people who produce or coordinate that aspect report to her.
Cloud computing has effectively separated information as an asset, from information systems as an asset. The value of the latter is now the subject of a considerable debate. At a time when an organization’s information systems (IS, as it used to be known) were exclusively comprised of data and the hardware that hosted it, it seemed logical enough for the people responsible for these systems to be certified technology professionals. Today, that argument is under fire. Some are arguing that the chief maintainer of information systems is no more deserving of executive status than the chef of the headquarters building’s commissary.
The more moderate approaches suggested for retooling the CIO role are nearly all, to some extent, demotions.
Executive talent search firm Russell Reynolds Associates advocates the creation of a Chief Digital Officer (CDO), which it describes as someone skilled with marketing who can take charge of the strategy for presenting the company’s information and its brand image to the customer. On a 3D value chain, think of a CDO as externally facing, with oversight over the company’s external roles. The CDO’s counterpart on the axis of internal roles might be described (using my phraseology, not RRA’s) as a “chief intelligence officer,” overseeing the strategy for information inside the organization. RRA’s ideal CDO is in clearly in charge of a revenue stream, as opposed to whoever has responsibility over the amortizable, intangible asset that is the company database. Writes RRA:
The CDO needs to be someone who not only has digital acumen but also is a seasoned general manager who can operate within a large-scale business and influence effectively across the organization. This is a relatively new type of leader and one who is hard to find, attract and retain. The increasing importance of the role eventually will put CDOs in the queue for CEO succession. CDOs will be the executives with the operating experience, management skills, strategic mindset and vision to lead businesses in an increasingly technological future.
Scott Brinker, the CTO and co-founder of Web marketing platform provider Ion Interactive, perceives a related concept he calls the chief marketing technologist. This person too has a purview that includes the outward-facing information from the organization, or what marketers call “the experience.” But unlike RRA’s ideal, which places the CDO on the proverbial chessboard as the queen beside the king, Brinker perceives this person as reporting to the existing CMO, who is technologically skilled, but who “choreographs” (his word) the digital marketing strategy as articulated by the CMO, for the CIO to carry out.
Forrester Research vice president and principal analyst Sheryl Pattek referred to Brinker’s ideal in a recent interview with me. We’re all familiar with the phenomenon of employees deciding on their own to bring consumer-geared technology into their organizations, including cloud services like Dropbox and Instagram. But in the companies she’s studied, Pattek has also seen the reverse: a tendency for the marketing division to spend money anyway on digital initiatives, especially in the cloud, when the IT department insists on adhering to traditional rules of procurement. It’s called rogue spend.
If I’m a CMO and I decide that I really need to enhance my mobile customer engagement strategy, and IT tells me it’s going to be nine months until they can get to me, do you think I’m gonna wait? I’m going to reallocate my existing marketing budget and find some dollars to go through my buying journey, and find the SaaS solution that will solve my problem, and I’m going to go do it on my own.
The IT organization has focused so much on infrastructure and maintaining those back-end systems, and the way they’re measured and incented is really around controlling costs and managing risk. Whereas on the marketing side, CMOs are really chartered with growing the business, and with trying to take risks to find better ways to engage customers and enhance the overall business operation. And those two tend to be diametrically opposed to each other.
Pattek recently co-authored an update to a Forrester report entitled, “The CMO’s Role in Technology Purchasing,” in which she shares new ideas for aligning (there’s a familiar word) the CMO with the CIO. It’s a problem that Accenture has also tackled in a public report called “The CMO-CIO disconnect: Bridging the gap to seize the digital opportunity” (PDF available here). Based on its 2012 survey of 400 senior marketing executives and 250 senior IT executives in 10 countries, from organizations without a chief marketing technologist (more on that distinction in a moment), Accenture states that roughly half of CMOs surveyed wish they could get around IT’s intervention in their decision making, and half of CIOs surveyed gripe about the marketing division acquiring new technologies without adherence to standards, compliance, and governance.
“This is the crux of the issue: who operates the technology to drive outcomes, who controls the design of experiences,” writes Accenture. The firm goes on:
Essentially, CMOs view the CIO organization as an execution and delivery arm, not as a driver of marketing strategy and excellence and a partner to be considered on equal footing. CMOs expect much quicker turnaround and higher quality from IT, with a greater degree of flexibility in responding to market requirements.
Resolving that crux is a simple matter, Accenture suggests: “Identify the CMO as the chief experience officer (CXO). CMOs must take responsibility for the consumer experience and drive consumer-centric measures.” Once those measures are identified, it’s then the business of the CIO, continues Accenture, to align itself with the CXO as a “strategic partner,” both of whom become “laser-focused on crafting the most relevant consumer experiences.” Though Accenture borrows one of its customer’s phrases for this org-chart relationship — “dotted-line reporting” — the cause-and-effect chain suggests the CXO is the one with 51 percent responsibility.
As Forrester’s Sheryl Pattek continues:
When IT and marketing are in this unique position to be able to turn technology into a business advantage, marketing people should be in a position to define the strategy, and the “what” and “why” of what they need, but they really should turn to their partners in IT to figure out the best way, the “how.”
You might expect to find an alternative viewpoint from a company like IBM, which has historically championed the CIO role in organizations. Back in 2009, at the risk of creating a Medusa-like image, IBM projected its vision of the CIO in mid-market businesses (PDF available here) as the principal driver of both business strategy and value.
But in a conversation with me for ReadWriteWeb back in February 2012, IBM’s VP for big data, Anjul Bhambhri, explicitly told me that as organizations build up their skill sets for interpreting the emerging data sets from piles upon piles of social media analytics. “I would claim, if we can make the CMO a data scientist,” Bhambri told me, “then we have achieved what we needed to, and that's when the potential of big data will actually be realized to get the business outcome that we think can come out of tapping into what's hidden in this big data.”
More recently, IBM’s Ric Telford followed up with me on this forecast. At the onset of the client/server era, he recalls, CIOs responded to the need for more options, more flexibility, and more functionality. But at the onset of the cloud era, IBM’s purchasing conversations are less and less with CIOs, he tells me, and instead with line-of-business professionals including CMOs. Says Telford:
What’s happening is, the buying behavior of the enterprise is changing, the dollars are shifting [with respect to] who controls what gets spent on IT in the enterprise. But in my opinion, the role of the CIO doesn’t go away. It evolves, and it really should evolve to what CIOs should be doing. Today, CIOs spend a lot of their time in crisis mode, infrastructure failures, outages, triage, and post-mortem problem determination.
You’d like to think that, over time, as data centers can get more efficient in the way they deploy through private and public cloud, the CIO can turn his or her attention more toward a more holistic view of what IT can do to move the business forward, and get a seat at the table — a kind of traffic cop, coordinator role, to help guide investments [with] the best options that are now available. That could be SaaS opportunities with industry verticals, or an infrastructure provider like IBM who can augment the infrastructure footprint of a data center, or a common platform for deploying new mobile applications.
Somebody has to guide all that. My biggest fear is that we’ll have client/server all over again, where all the lines of business will often buy their own systems, and then the company [departments] can’t talk to one another, and spend most of their time doing integration because they didn’t have a coordinated strategy.
Telford goes on to tell me a story about how decades ago, he spearheaded the development of one of IBM’s first client/server calendaring systems. That system, he recalls, had to support five separate network protocols, because at that time, the various departments of IBM’s corporate customers had each taken it upon itself to be the driver of company value. In so doing, they had adopted competing protocols such as Microsoft Windows NT, NetBIOS, Novell NetWare, TCP/IP, and IBM’s own SNA, without feeling they needed to consult with one another first. They barely shared lunch together, let alone collaborate.
“It was chaos,” he says. “Every application had to support all these different networking protocols, because everybody had made different decisions about what they were going to implement in different parts of the business.”
This was before the CIO role was created to coordinate purchasing decisions, and effectively drive a single track for technology support. It made clear business sense in the 1990s and 2000s to converge each business silo’s technology track around one network and one platform, and for one department to oversee that platform. But now that infrastructure can be sliced and diced, and five or six virtual infrastructures could cost as much as one big one five or six times the size, the cost-cutting scalpel is being sharpened for use on the IT department itself.
“E”
The driver of business value in the 21st century organization has become a jump ball. Making the strongest argument in its own favor, naturally, is the department in charge of making strong arguments: marketing. Maybe the customer is the true determinant of value in a business relationship, marketing now concedes. If that’s so, maybe marketing should be the one providing the customer with the information and the positive brand image she needs to be able to choose the company as her supplier. And in that role, marketing should be the one choosing the technology it implements. This argument appears to be just as valid for B2B relationships as for retail.
While the IT department is busy crafting a counter-argument, marketing is going ahead and making purchases. It’s the elasticity of modern virtual infrastructure that makes this possible. The eagerness of marketing to make these changes immediately, perhaps before everyone else takes notice, is testament to how little alignment there actually was between the CMO and CIO over the past quarter-century.
When a C-suite officer is given purchasing authority — or, as may often be the case, a C-suite officer seizes it — she acquires with it the reins of authority over the people who implement her itinerary. That was the intent behind the creation of the CIO role in the first place: centralizing both the purchasing and the managing authority that, during the client/server era, had become subdivided among several silos, rendering them corporations unto themselves, making them compete with one another instead of collaborate.
Lurking in the background of all this, though — perhaps under the rug — is a new fact that we tend to bring up in conversation without fully interpreting its meaning: In the map of the “customer journey,” to borrow Sheryl Pattek’s phrase, the customer is at least three-fourths of the way towards a complete comprehension of a product’s or service’s value, as well as the brand behind it, before he even introduces himself to a prospective supplier. The “cold call” is dead.
It’s nice to pretend we’re still in the 20th century, and that any branded company can still be one-hundred-percent responsible for forging the “customer experience.” Perhaps the last vestige of the Web portal era is the flickering hope that the customer experience can be programmed, debugged, and played back for each customer, accompanied by popcorn, sodas, and Milk Duds. Yes, the company is responsible for the brand image it presents to the world. But the world is no longer a silo, just as the data center is no longer a silo. It’s a network, and the customer’s experience of that network is two-way. Not only does a customer experience the image that the brand conveys of itself, but also the images presented by other customers.
The power of social media, you see, is quantified more often than it is identified. Social media is a conveyor belt for influences. It is the growing influence of the consumer that is tipping the scales the other direction throughout the economy.
The phenomenon referred to as the “consumerization of IT” is really just one symptom of a much larger trend: the growing power of the individual to influence business relationships. Even amid the diminishing signal-to-noise ratio that afflicts the Web, the individual still has greater access to information than ever before.
So where’s the “chief experience officer” for managing the feedback? What plans should a company make for responding to the influence of the consumer, the buyer, the prospective partner? The first generation of SaaS platforms for interpreting outside influence are focused on containment, damage control, even drowning out the negative messages with positive ones. Their objective has been to neutralize the force of customer influence, and their effects on markets thus far can only be described as spectacular backfires.
If information output should have its own champion in the C-suite, then it only follows that information input should have its counterpart as well. Customer relationship is at the core of the first SaaS platforms for managing this evolved communications process. If it can be leveraged for marketing, it should also be leveraged at the opposite end for intelligence. The platforms do not need to be split. And one chief-something-officer should not find himself reporting to the chief-something-else-officer simply because that other person bought the platform to begin with.
In an organization that gives equal weight to its internal and external roles, the ideal state is what I call equilibrium. Marketing drives the customer perception of quality, while technology drives the business perception of margin. Both combine to form the customer-driven vision of value. The trick here is to employ someone who can oversee the balancing act, who can make sure neither side clamors for that 51 percent status. That’s what I thought the “E” was all about in “CEO.”
