I don't usually do much commentary on industry happenings these days, but things do seem to be heating up as of late.
It's getting interesting again, especially with regards to all things "cloud".
I had a good introduction to the Hindu pantheon at an early age, and I keep thinking of cloud as Shiva The Destroyer -- knocking down the old IT world order, and creating the opportunity for the next.
Simply put, cloud changes all three aspects of traditional IT: the technology model, the operational model and the consumption model.
Everything we know is being remade, and will be remade again and again.
Anyone who claims to have this all figured out is either being presumptive, or is far smarter than I.
You can appreciate my fascination :)
Private Clouds Are Here To Stay -- And Apparently Growing
One of the more vitriolic debates amongst the clouderati has historically been the role of private clouds.
There were many who just couldn't grasp why IT organizations weren't getting out of the business of owning IT assets, and instead consume 100% of their IT needs externally.
For those of us who've spent their career working with enterprise IT organizations, we know the reality: there are very strong structural motivations for private clouds: economics, control, etc. -- and those aren't going to change anytime soon.
I'd like to be able to point to some reasonable industry analysis that backs up my assertions, but I can't. Since there is no generally accepted model for what is a private cloud (and what isn't), we're left with reading tea leaves that act as real proxies for the action going on here. My informal sampling of customer groups is right around 90%: just about everyone has a part of their environment they refer to as their "private cloud" or similar.
One indicator might be virtualization adoption. Hypervisor technology within the enterprise is now almost ubiquitous. I even meet customer groups that are somewhat embarrassed that they're "only" 80% or 90% virtualized. Who would have believed that just a few short years ago?
Another proxy might be purpose-built infrastructure. VCE appears to be setting the pace for data-center-class converged infrastructure (now a >$1B business) with both HP and IBM hot on their heels. Moving down a bit, the reference-architecture builds (FlexPod, VSPEX, et. al.) are now a mainstay of the industry, routinely called out during earnings calls. Moving down even further, the pre-fab node model continues to be popular to many (think Nutanix, SimpliVity and perhaps the oft-rumored Marvin?).
All can be considered reasonable proxies for private cloud adoption. And, collectively, they represent many billions of dollars of IT spend, and appear to be growing quickly.
But there are better proxies: automation software for example.
For me, cloud is primarily an operational model, and that can't be achieved without a non-trivial investment in new workflows and tools.
I was pleased to see a recent IDC report that stated the data center automation software market grew an impressive 22% from 2012 to 2013, reaching a healthy $1.8 billion. I think it's safe to presume that most of that is going towards automating newer, cloud-like environments, e.g. private clouds.
Unsurprisingly, VMware emerged victorious with a #1 spot of 24.7% market share and an impressive 65% growth rate. That's momentum. However, a quick look will tell you it's still a very fragmented market indeed, with the next three vendors (IBM, BMC, HP) responsible for around 35% of the market.
Much Ado About Amazon Et. Al.
While there are some truly good industry journalists out there, some of them swim in the shallow sensationalist end of the pool. A recent example was Amazon's offering of a vCenter plug-in to enable easier provisioning of AWS services using in-context tools.
You would have thought a simple piece of plug-in software represented a mortal blow to VMware, given some of the coverage.
As many readers will know, provisioning on AWS and moving a handful of simple VMs across has long been relatively straightforward. The bigger challenge has been with the other 98% of the model: monitoring, moving stuff back, etc. -- all the gritty operational stuff that no one really understands until they've had to deal with it.
And nothing has really changed there.
By comparison, the defining attribute of VMware's vCHS (vCloud Hybrid Service) is that all-so-important operational compatibility. If you're running VMware tools, it runs as an extension of your existing environment.
Avoiding the over-used word "seamless", it's pretty darn close.
I've seen customer reaction as uniformly positive to the existing vCHS offering, and there's a boatload of cool enhancements in the roadmap if you can get someone to share it with you.
The public cloud business is the essentially same as any other: understand your customer and what they really need. The vCHS team seems to have done that.
One of the unverified memes floating around inside VMware is illustrative: if we can get 10% of our installed base to use vCHS in some fashion, it will be the industry's second-largest public cloud.
By comparison, my compatriots at HP have announced a new cloud strategy, one that appears to be challenged from the outset.
Part one is a massive and public investment in OpenStack. That's a good thing. Part two is to create an alliance of independent IT service providers to use HP's modified OpenStack software tech (Helion) to build their own services, often referred to as the "arms dealer" model.
That one isn't so good -- and I speak from direct experience.
When I was at EMC, we had a similar game plan. The first incarnation (VMware's VSPP) was only modestly successful in achieving its original goals. There were a number of issues that were difficult to overcome, but the biggest one from my perspective was that every service provider wanted to change the recipe.
Maybe they thought they could do better on their own, maybe they thought they had to differentiate from all the other services, or something similar. Some have become successful (CSC's BizCloud comes to mind), others less so.
During that period of time, I could see that customer reaction wasn't strongly positive. Yes, all the service provider partners were using essentially the same tech, but the resulant customer experiences were all over the map.
Say what you will about McDonalds, but when you walk into one you know exactly what you're getting. That's what customers have told me they wanted: consistency and compatibility of experience -- wherever they go.
I wish HP the best of luck.
Where Does OpenStack Go From Here?
Before anyone feels obligated to flame me, I'm a big OpenStack fan. My company -- VMware -- sees OpenStack as a clear opportunity to sell their software tech using a different integration framework. My previous company -- EMC -- has been genuinely enthused by the platform and the interest.
That being said, few would take the position that OpenStack-based products are ready for prime time. Not many IT organizations are prepared to invest the engineering muscle required to make OpenStack production-capable.
The current situation is analogous to Linux in its early days: everyone can see the potential, but it's just not soup yet.
Two things to keep in mind as we consider the proverbial OpenStack "tipping point".
First, the pressure on IT organizations to cloud-ify their environments is nothing short of intense these days. People want results, and they want them sooner than later. And few can afford to delay heavy investments in tech and process waiting for OpenStack-based products to mature. Once their environment is operational (hopefully with a majority of VMware products), it's tough to make a case to come back and do it all over again anytime soon.
Second, we're likely going to see far more "productization" of OpenStack code in the near future -- mixes of open source and proprietary software, with the usual constraints on what's supported, and what's not.
Building software products costs real money for real engineers, which will inevitably negate some of the perceived economic (and "open") advantages some are hoping for today.
Will we see more OpenStack-based private clouds in enterprises in the future? Bet on it. But it's going to take longer -- and perhaps be more moderate -- than most people expect.
And On To PaaS
It was a distinct pleasure to follow the recent Cloud Foundry Summit online. A thousand people showed up -- folks who are passionate about what CF can do for them today, and in the future. I found many of the stories impressive. And the momentum appears to be accelerating.
The big idea? Perhaps the most important thing PaaS (and Cloud Foundry) does is abstract application from infrastructure in a meaningful and useful way. Developer productivity accelerates, significant applications can be deployed far more quickly, and there's far less dependency on the specifics of the underlying infrastructure.
In a post-cloud / multi-cloud world, it's the next frontier.
Not everyone does significant application development, but many do. And, as business models change in a digital world, more and more organizations will find themselves in the software development business.
That means considering a strategic investment in your software "factory".
From my biased perspective, it seems that -- right now -- it boils down to two choices for modern enterprise PaaS: Cloud Foundry ... or -- ??? -- well, I really don't know. The Cloud Foundry Foundation strikes me as an elegant governance model to ensure the direction remains aligned with stakeholders, and it's garnered substantial support (34 members making sizable contributions) in its first few months.
What I really liked about following the event was that there were plenty of real-world examples of the tech being used today to deliver eye-popping results. Let's hope the strong momentum continues.
It's Still Early Days In Cloud
There are plenty of commentators who are willing to claim "game over" when it comes to cloud.
Horsefeathers. It's as short-sighted as claiming "game over" for the internet, mobility, etc.
Just when you think you're beginning to understand something, it morphs and begins again.
We're in the opening chapters of what will prove to be a long and exceptionally interesting story.
------------------------------------
Like this post? Why not subscribe via email?
(and a HUGE thanks to the many hundreds of you who already subscribe!)
Comments