Ours can be an industry of big ideas.
Certain ideas can quickly transform the landscape much the way a summer thunderstorm can quickly form and unleash powerful forces around it.
Such appears to be the case with "private clouds".
Starting as an obscure concept introduced at the beginning of this year, it now shows the promise of becoming not only the dominant model for next-gen enterprise IT, but perhaps service providers as well.
While no one in this industry is a definitive authority on anything, I do have the pleasure of being at "ground zero" as this concept was formed and communicated.
And exactly why this concept is becoming so popular is a story in itself.
In The Beginning
Towards the end of 2008, VMware, Cisco and EMC realized that we all had a shared view of how enterprise IT was likely to evolve.
Virtualization would continue apace, and enable entirely new operational models for IT. Service providers would enter the marketplace with compatible offerings based on virtualization.
And enterprise IT organizations would have entirely new options about how they consumed IT infrastructure.
We felt so strongly about this shared vision that we took the step of forming a small team to formalize the share concepts, and start using the same words and pictures to communicate with our various stakeholders.
This shared effort was only one of many shared efforts we were starting to do in the context of our partnership.
What we needed early on was a generic industry term to define the category we were thinking about.
If you've ever tried to name something -- and get everyone to agree! -- you'll realize it's a bit more involved than throwing names against a wall and seeing what sticks.
We knew that we wanted the word "cloud" somewhere in the term -- the environment we were considering had many cloud-like attributes -- elasticity, geographic independence, etc. -- but was a specific kind of cloud, hence the need for a modifier to go with it.
At that time, most enterprise IT groups weren't enthusiastic about most of the cloud offerings being bandied about from Amazon, Microsoft, Google and others.
Most of these offers presumed that applications would be re-written or modified to run in a cloud. Not good, from an enterprise IT point of view.
Few, if any, offered IT any control over how the resources got used: service delivery, security, policy compliance, etc. Most enterprise IT activities require that enterprise IT be in control of what goes on.
And, finally, interoperability choices were few -- there was no talk at the time around federating internal and external resources together, or interoperability between service providers. Also a non-starter for most enterprise IT groups.
We finally settled on the word "private" to describe the kind of cloud we were envisioning -- private clouds were different in that (a) they didn't require applications to be re-written, (b) they used any combination of internal and/or external resources, and (c) there would be an open marketplace of compatible service providers to offer enterprise customers choices.
At the time, the term "private cloud" was not in general use as an industry model. There were a few smaller vendors using the term to describe a proprietary offering, and a few insightful analysts probing around at the concept, but -- generally speaking -- not much out there.
Update: @sfoskett tweeted that I was somehow claiming credit for coming up with the term. Not true -- I later found some great work done by Thomas Bittman at Gartner using the term "private cloud" written a few months earlier in 2008. If anyone should get credit for originally coming up with the term, it'd probably be him, and not me. Since the "who was first" debate can consume an inordinate amount of time, feel free to offer up any examples prior to Sept 2008.
We Start Telling The Story
The first target audience was uber-large IT organizations who had a serious effort underway to re-define their IT infrastructure strategy. We wanted to see how our ideas stacked up against other concepts and approaches out there.
Frankly speaking, we were a bit surprised by the strong positive reaction. Yes, there were questions and concerns, but -- generally speaking -- these large customers liked the architectural approach, and liked the fact that three industry leaders were working together to make it happen sooner than later.
Before too long, we became backlogged with requests to have these sorts of customer meetings, and we got busy training more people who could tell the story and engage with customers. Despite our best efforts, we still haven't been able to keep up with demand to this day.
The Cloud Community Weighs In
In one regard, we were entering an ongoing cloud discussion from an entirely new perspective. Understandably, there were reactions from people already immersed in this stuff.
One early reaction was "you can't call it a cloud if it involves assets inside the data center" or something similar, the idea being that clouds could only be called clouds if they were completely external to IT.
We saw that as only one potential consumption option in a much broader set of choices: use external resources, use internal resources -- or any dynamic combination. It turned out it was hard to argue with that kind of logic.
Another reaction was "you can't call it a cloud unless applications have been re-architected and optimized to take advantage of cloud services". Again, there's nothing wrong with rewriting an application for a particular deployment model, but we felt that shouldn't be the price of entry.
Indeed, we could clearly see many compelling cloud-like advantages that could be gained by simply encapsulating existing server and desktop applications in virtual machines, and orchestrating at that level rather than forcing it on the application.
Now that I think about it, there were many people who perhaps thought cloud had to be difficult and revolutionary to be a meaningful topic. Confronted with a view that showed a path to cloud benefits that was comparatively evolutionary and non-disruptive brought out a fair amount of emotion.
Over time, these discussions died down, and you rarely hear them anymore.
It's Not Just Big Enterprises -- Duh!
As time went on, we started sharing these discussions with audiences other than uber-large IT strategy groups, and were surprised by the positive responses we saw from these groups.
We ended up talking to IT organizations large and small, and -- generally speaking -- if they were committed to a virtualization strategy, it seemed the most logical thing to extend notions of fully virtualized environments to private clouds.
We started to talk to our partner community: resellers, integrators and the like. We were surprised to find out that -- generally speaking -- EMC partners were also VMware partners and likely Cisco partners as well. Needless to say, they really liked the story, and especially the fact that the three companies were working closely together to accelerate the vision.
As of this writing, the private cloud story is our preferred strategic vision for most any IT infrastructure audience.
Formal Outings Begin
It's one thing to be sharing these thoughts in small audiences, it's another thing entirely when CEOs take the big stage and do the same.
The first such public outing was done by Joe Tucci and Paul Maritz together at the EMC Strategic Forum last March in front of hundreds of analysts and press.
Paul Maritz and the VMware team repeated the theme at VMworld in Cannes, as well as during the vSphere announcement earlier this year.
EMC repeated the them as part of the Symmetrix V-Max announcement, and later the Ionix resource management announcement -- not to mention heavy coverage at EMC World.
And Cisco used these themes as part of their UCS announcement, and later from the stage at Padmasree Warrior's Cisco Live event.
There's more, but you can see what's already starting to happen -- all three companies are lining up behind a shared vision.
Customers Want To Take The Next Step
Before too long in all this, we started meeting customers who wanted to go from vision to action.
The road ahead for these customers generally evolved into a couple of coherent themes.
One dominant theme was needing help in virtualizing tier 1 applications (e.g. the ones that could get you fired). These customers had successfully virtualized most non-critical applications, but were having trouble tackling the big ones.
We started using our EMC Proven Solutions capabilities -- coupled with professional services -- in an effort to assist with this. We could show exactly how to do this, and what you could expect if you did. Eventually, though, the challenge turned out to be less around technology, and more around changing mindsets -- a theme I'll come back to in a while.
Another dominant theme was optimized infrastructure -- making sure the technologies that they were deploying would not only solve today's challenges, but serve them well in fully virtualized environments leading to private clouds. We started changing our product presentations to reflect these themes.
A special discussion emerged around management and security disciplines. It was easy for us to see that these capabilities could be far better in the virtual world than the physical world, but to get these advantages, customers were going to have to change how they did things.
We needed to be upfront with people that learning to manage and secure IT infrastructure in new ways was not going to be easy, but was going to be essential to gain not only the potential resource efficiencies, but the operational efficiencies as well.
A more interesting set of themes evolved around next-gen operational models and policies. Yes, the newer tools were capable of all sorts of cool things, but IT's operational models and associated policies had evolved in the physical one, and not the virtual one This turned into a consulting-style relationship to not only identify these new models and policies, but to identify change management techniques to move from legacy thinking to newer thinking.
And a surprisingly large number of customers wanted to shortcut all the people, process and policy issues using an entirely different approach: stand up a baby private cloud, move some sample workloads over to it, get comfortable with the new processes, workflows and policies, and measure the results.
The idea was to start with something new, move some representative applications to it, and gradually expand the bounds of confidence without bringing forward legacy technology, tools, processes and policies.
Of course, that sort of thinking really gets me excited. It's a blast to meet customers who want to do this sooner than later.
The Federal Government Responds
The other gratifying response came from various US government agencies.
They liked the idea of evolving what they had (rather than starting over again). They liked the idea of an open marketplace of compatible service providers, including running in their own data center. And, since timelines tend to be long in this space, I think they liked the appeal of a simple idea that could be uniformly communicated, and no need to wait for new technology to be invented.
The Industry Responds
You know you're having an impact when you see your competitors starting to respond, either validating your ideas or sometimes the exact words you've chosen.
Amazon, for example, has now started to offer virtual machines that don't require using all of Amazon's stack. They're also talking more about offering IT shops more control around more aspects of how the resources get used. All steps in the right direction.
At the recent Microsoft conference, Steve Ballmer and others started talking about "private clouds" and the need to offer customers choices as to whether workloads run internally, externally or any combination. No specific offerings yet, but they're thinking in the right direction, I'd offer.
IBM did something very clever indeed. Previously, they had been positioning cloud-for-enterprises as a variation of outsourcing -- move extra workloads to an IBM-branded outsourcing facility. Not exactly what most customers were looking for.
Their new approach was to deliver a "private cloud in a box" -- servers, storage, VMware and a bit of management software -- and target a very specific use case: test and development. Lots of missing pieces in their story, but you can see how they're working towards the broader concept, albeit from a different direction.
Oracle offered up a unique, Oracle-centric view of "private clouds" -- talking about offering to manage Oracle environments in customer data centers on behalf of their customers, and being able to do the same in their managed data centers. Interesting, but I think most customers will want to choose who controls their environment, and will probably want more choices than A or B in terms of infrastructure models. Not to mention that their approach is -- of course -- limited to Oracle-based applications.
HP might be doing something in this space, but I haven't heard anything definitive. One can only wonder what the Oracle/Sun combination will produce, but there might be an interesting model emerging from this combination.
The pure storage plays (3Par, NetApp, etc.) are starting to talk about how their storage will play in service provider clouds that look similar to the type of private clouds we're discussing here.
Sure enough, there's plenty of evidence that more and more IT infrastructure players are starting to line up behind the concepts, which is a good thing. If you don't believe me, feel free to search for "private cloud" on Twitter, Google, etc.
Even Service Providers And Outsourcers Are Responding
You might be tempted to think that people who rent infrastructure as a service might be farther along than the customers they're targeting. There are plenty of large players, but we haven't seen many announcements -- yet! -- from some of the more familiar name brands in this space.
Part of the challenge is nailing the business model. To these people, it's less about the technological glitz, and more about offering, pricing, go-to-market, synergies and competitors. It's taken a while to ramp up these discussions, but I think you'll be seeing far more public engagement in the near future.
And There's A Bunch You Can't See As Well
The private cloud story is compelling enough in its own right.
But -- for me -- the exciting part of the story is how the three companies are working together in some innovative ways to make this all happen sooner than later.
Right now, there's a ton of activity going on at VMware, Cisco and EMC to line things up -- product roadmaps, integration efforts, solutioneering, support, customer engagement, partner engagement, etc. -- all with the same goal: to help customers and partners transition to fully virtualized environments (and private clouds) sooner than later.
I've never seen anything like it before. It's pretty exciting.
The specific details (and they are considerable) will have to wait for a formal announcement at some point, but there is already plenty to talk about, with more coming at a regular cadence.
And, of course, we all have our own ecosystems of partners we work with that may or may not be friendly to the others -- this is not an exclusive relationship by any means. But, frankly speaking, this isn't your usual alliance relationship that's typically based on powerpoint and little else of substance.
So, Where Do We Go From Here?
First, I'll continue to blog on this as one of my major themes. Not only is it related to my role at EMC, but it's one of the more fascinating industry transformations I've ever been associated with, and there's so much to talk about.
Just as an example, here are some of the posts I've offered up over the last few months:
- some relatively clear thinking about clouds, around three key aspects: architecture (how they're built), operations (how they're run) and consumption (how they're paid for).
- trying to identify big policy issues that will undoubtedly get in the way, such as getting comfortable with oversubscription as well as other barriers to adoption.
- getting people to think about desktop experiences as an integral part of the private cloud.
- Lots of discussion for telcos and service providers, not only identifying the opportunity, but a few prelim discussions around business models.
- Opening discussions around traditional disciplines such as management, backup, business continuity and security.
- A few experiences I've had with customers who are actually building these things.
- Some generic reaction from press and analysts, for those who follow that thread.
You might think this is a lot of material for a single topic. These are just highlights -- you can see all my related posts here. The scary part is that there's at least several dozen posts to be written -- not only on these themes, but other related ones.
Second, maybe you've seen what's being referred to as the "private cloud / VCE" material. If you have, I'd be interested in your reactions. If not, we'll have to find a way to get the story to you in such a way that you can engage and react. The good news is that our collective bandwidth for doing so is increasingly dramatically.
In particular, if you have some time, one of the more popular items is this 30 minute video from last April of me outlining the whole story. Warning: 140 MB, wmv format.
Third, I hope you'll be interested in hearing more -- not only from VMware, Cisco and EMC -- but from other vendors and ecosystem players. I'm betting that the industry-wide story lines up and becomes widely compelling before the end of this year -- although fiercely argued at the detail level!
The party has just started ...