Economic bubbles are formed when inflated expectations greatly exceed reality. In tech, we see the same effect: a new idea storms the market, expectations greatly exceed reality, and reality comes crashing down hard.
I would argue that we're now in the aftermath of the first public cloud bubble bursting. There may be more cloud bubbles ahead of us. Simply review the last three or so years, and you'll see all the signs: greatly inflated expectations driving substantial investment, followed by demand failing to materialize, followed by a sharp correction when reality strikes.
Understanding what worked -- and what didn't -- can be instructive. Unless we learn from history, we're doomed to repeat it. And I'm betting this won't be the last cloud bubble we'll see.
So what can we learn from the first cloud bubble?
I trace the popular origins of the first cloud bubble to Nicholas Carr's landmark tome The Big Switch.
Drawing strong parallels to the transformation of the electricity industry during the early 1900s, he made his case that the the IT landscape would transform to a small number of giant utility cloud providers for similar structural reasons: economies of scale, specialization of roles, etc.
The power of Carr's arguments -- plus the rapid growth of a much smaller and less pervasive Amazon AWS as tangible evidence -- convinced many IT vendors to believe that significant disruption was at hand, and urgent action was required.
Cloud was coming, so we better build one -- fast!!
It looked like a herd mentality set in: just about every large enterprise IT vendor (HP, IBM, Dell, Oracle, Microsoft, et. al.) announced their plans to build multi-beelyun-dollar public clouds. Less visible were dozens of telco operators who bet on the same opportunity. More than a few SIs and outsourcers got into the game. And, behind them, literally hundreds of regional and specialized providers.
From my perch inside a large IT vendor, there was a period of time where it seemed that everybody and his brother wanted to build some form of public cloud service. But very few of these panned out, it appears.
Amazon AWS continues to do well, but not in the enterprise. The success of the larger IT vendors is very debatable. The telco clouds -- as a group -- haven't lived up to initial expectations. And, outside a few exceptional regional or specialized providers, the smaller clouds haven't thrived either.
Lessons So Far?
#1 -- Clouds Thrive When They Satisfy Unmet Needs
To me, Amazon AWS is an excellent example of this. Prior to Amazon, there was no off-the-shelf IT infrastructure service that was easy and flexible to consume, variably priced, eminently controllable by end users, etc. Largely speaking, their best customers had little or no legacy to deal with; so adopting Amazon's way of doing things was a perfectly reasonable constraint.
If I were to (once again) use the tired car analogy, Amazon offered a convenient car rental service at a time when the only alternative was to either build your own car, or pay someone else to build it. The model attracted a vast number of mostly non-enterprise customers whose needs couldn't easily be satisfied by other needs. Good for them.
Look a bit deeper, there are many dozens of successful enterprise-oriented SaaS-ish and PaaS-ish clouds that offer attractive alternatives to purchasing and assembling the pieces yourself, and they do well. I would expect this to continue.
For potential cloud vendors, the rubric shouldn't be "how do I compete with Amazon or Rackspace or ...", it should be "what important unmet customer needs can I satisfy using a cloud model?"
#2 -- Clouds Can't Mismatch With Enterprise IT
I can't tell you how many times a cloud service provider had told me that they plan to compete solely by being less expensive than in-house alternatives. You'd start asking some hard questions around difficult enterprise requirements, and they'd come back to the "well, we're cheaper" as if this was some sort of magic trump card.
I couldn't see this ending well.
Satisfying key business requirements is Job #1 for enterprise IT; doing so for less money is somewhat secondary. Business requirements include things like: predictable performance, security, resiliency, recoverability, compatibility with existing applications, being able to resolve problems quickly, compatibility with existing workflows, passing an audit, etc.
If your cloud service doesn't have good answers for one or more of these topics, it's not about price -- it's about being fit for purpose.
Going further: enterprise IT has come to expect a high-touch engagement. That implies a raft of skilled people as part of the enterprise cloud business model: sales reps, technical presales, solution architects, customer service, etc. -- and that's not an inexpensive proposition.
As an IT vendor, you're known by the quality of your people; I can't seeing it being all that different in an enterprise public cloud world.
Simply expecting enterprise IT organizations to "figure it out for themselves" from your website isn't reasonable.
#3 -- Cloud Is Really An Operational Model
At some level, anyone with sufficient resources and skills can buy and operate a cloud service. That includes enterprise IT organizations of reasonable size.
They run the numbers, and inevitably realize that they could potentially deliver better (and cheaper) IT services internally than going outside -- especially for the day-in, day-out predictable workloads that comprise so much of enterprise IT.
That realization is one of the motivations behind ITaaS and IT transformation: the recognition that -- at some level -- cloud is nothing more than an operational model for producing and consuming IT services, and it can often be replicated within the enterprise -- hence the strong and growing popularity of private clouds.
#4 -- Cloud As A Key Skills Arbitrage
Running an enterprise-grade cloud service requires a critical mass of unique and somewhat scarce skills: working together as a team across a variety of disciplines: architecture, operations, security, capacity planning, etc.
Putting together such a team is no easy feat; one that can often be beyond the reach of even good-sized enterprise IT organizations.
I always wonder what would have happened if all those less-than-successful cloud providers *didn't* focus on technology and *didn't* focus on being cheaper, but instead focused on their team: skills, roles, processes and expertise. Why? There's ample precedent for enterprise IT going outside for expertise -- even if it's delivered in the form of a service.
The pitch then becomes "we've got a team of exceptional people who can deliver critical IT services faster/better/cheaper than your own team" -- and that's hard to argue against in many situations.
From The Ashes?
If I made a list of all the cloud services that have been launched -- and failed to gain traction -- it would be quite a list. Yes -- there have been successes, but they have been the exception, and not the rule.
It ain't over yet. The notion of enterprise-grade public clouds isn't dead, it's just taking a while to mature. Expectations have been reset, and we move forward.
I would expect over the next few years we'll see far more "enterprise-friendly" cloud services, ones built on sustainable value propositions vs. ephemeral considerations.
First, I would expect to see more cloud services that mirror enterprise IT: familiar technology, familiar operations models, compatible control planes for important functions, and so on. Public clouds that don't force enterprise IT to compromise or redefine their model just to consume a service.
Second, I would expect to see newer enterprise-oriented cloud models that target parts of the landscape that are (a) important to the business, and (b) somewhat hard to do. The proposition here isn't necessarily doing things cheaper; it's getting results faster.
Potentially falling into this category would be things like big data analytics, sophisticated application development platforms, advanced security, collaboration, mobilization, etc. A big requirement hits IT, and they can't mobilize fast enough to meet the requirement, so they go looking for compatible external services.
But, once again, unless these services are aligned with the enterprise IT consumption model, they too will struggle.
Third, I expect that these newer enterprise cloud services will recognize the inevitability of the required high-touch model: investing in smart people that face directly with customers, and help them through the entire engagement.
Finally, I would expect complete transparency to be a fundamental business tenant: sharing each and every aspect of how the service is implemented, delivered and supported -- warts and all -- which is most decidedly not the norm today. I can't tell you how many cloud providers insist that "customers don't need to know the details".
Yes, they do.
Enterprise IT customers routinely demand complete transparency from their product suppliers -- why would they expect anything less from their cloud service providers?
Good Things Take Time
If you think about it, we're talking about two very important and hard-to-achieve evolutions that have to happen at the same time.
The first are the emergence of a rich marketplace of enterprise-grade clouds (and engagement models) that align well with enterprise needs. There are some today, but not nearly enough.
The second is the evolution of the enterprise IT organization itself: from an implementer of technology to a business-oriented service broker. These days, the progressive IT organization sees itself as a business-oriented "service provider of choice" that builds or brokers services as makes sense.
And not every IT organization has the ability -- or the justification -- to invest in learning how to operate an enterprise-grade private cloud, or a big data analytics environment, or a fully mobilized collaboration environments, etc.
Put differently, demand for enterprise-grade external clouds appears to be increasing dramatically - simply because more IT organizations are evolving to a point where they have clear needs.
And there's nothing like strong customer demand to stimulate a market.
Like this post? Why not subscribe via email?