Recently Randy Bias, Simon Wardley and David Linthicum wrote excellent articles on what OpenStack needs to be successful and what the likelihood of success is. Each of the articles made great points and are well worth the read, with many spot on arguments. I point out in each case a few examples of where I disagree or agree and why.
Smart & Experienced Guys
I know Randy, Simon and David and each of them has the kind of knowledge, intelligence and experience that justifies their leadership position in the cloud and IT space, so why am I “attempting” to argue with them? Good question, maybe I haven’t had a good beating recently.
To quote my friend Joe Weinman “I try not to get into arguments with people that are smarter than me” and yet here I am, stupid is as stupid does?
I hope I can at least bring a different angle to the discussion, even if it’s not a strictly technical angle.
My angle on the blogs from Randy, Simon and David
Randy suggests that better application design, combined with compatibility between private and public clouds is needed. The fact that I support this point of view shouldn’t be a surprise to anyone that knows me, as I’ve been a vocal supporter of hybrid environments and multi-cloud capabilities for many years. Where I differ is as much a matter of timing that anything else. I see the current phase of OpenStack adoption as serious exploration and POC. As such, real “use” versus “evaluate” is still 18-24 months away. By the time we get to real use, we’re likely to have better tools and strategies for OpenStack integration across distributions and public cloud stacks.
Simon says that OpenStack is effectively dead, and it just doesn’t know it yet. I heartily disagree and I base my disagreement on several factors of adoption and human nature, which I explain in greater detail in subsequent paragraphs. He also suggests that choosing an alternative path to cloud just because it’s not Amazon is wrong, a sentiment that I totally agree with.
David talks about the lack of cohesion even between distributions of OpenStack, and I agree with his view. I just don’t see the divergent distributions being a major hindrance to adoption for the reasons I cover in my comments for Randy and Simon, along with what I discuss below.
My take on OpenStack development and in very simple terms why it’s likely to succeed
Private cloud (OpenStack) will succeed because companies and people need it and want it to. The forces at work here are far beyond the capability of Amazon (AMZN), Google (GOOG), and or Microsoft (MSFT) to battle. Imagine a potential discussion in HP, IBM, or Cisco’s boardroom “It looks like Amazon & Google are going to be doing all the cloud, I guess we should just give up on server and storage sales for good”. The conversation about how to stay in the game is happening, and it’s happening around how to ensure the customer gets a solution that’s good enough. The solution will be easy to install (converged infrastructure, with a xStack) so that it helps the customer forget the notion that Amazon, Microsoft or Google are the only real cloud choices. Imagine the disruption to the rest of the ecosystem if all hardware ends up being built specifically and only for the large public cloud providers. In other words, there is a giant vested interest in the traditional IT vendor community to make a solution that will keep the enterprise customer happy and on track to keeping effective and efficient IT infrastructure (in part) internal.
On the customer side, there continues to be the assumption that we should build it and own it. While this assumption often gets IT into trouble, in many cases it’s also the only thing that saves them. After all, Google and Facebook (FB) are nothing but big IT shops that have decided they needed to build it themselves. The determination in the enterprise to hang on to the ability to design, build, and manage highly complex infrastructure means the efforts of companies like HP (HPQ), IBM (IBM), Cisco (CSCO), and others won’t be in vain. The trick is primarily whether or not they can deliver infrastructure and cloud platforms (I.e., OpenStack, vCloud and CloudStack) in ways that make it simple for companies to install and put to use. So far, I’d say they are being successful on the converged infrastructure portion (vBlocks, Flexpods, VRTX, Matrix, UCS, etc). One of the biggest pieces still missing is that hole that OpenStack is supposed to fill. Yes, right now it’s still a .XX (.80?) version, but rarely in history has there been more money and support behind a project in the IT space than there is for OpenStack right now.
Risks that could ruin the future for OpenStack, CloudStack and private cloud in general
IT organization and business integration
- If IT organizations can’t start looking beyond the technology of cloud and thinking more about ITaaS, organizational agility and business integration, then private cloud doesn’t stand a snowballs chance in hell of surviving.
Rough feature parity with public cloud offerings
- There need to be improved tools for making internal cloud look, act, and work in similar simple ways to public cloud. The features and capabilities don’t have to be a one for one match, but they need to be closer than they are today. Without rough feature equivalency, internal cloud installs will struggle to keep developers/users happy.
Risks that likely won’t change the outcome of what I’m arguing
Divergent distributions of OpenStack:
- I’ve been saying since about 2011 that the future of OpenStack is likely to look much like the Linux market with unique distributions that are developed and supported by key suppliers. I don’t see this as a major inhibitor to adoption. The fact is many enterprises are diving in knowing they don’t have a fully supported (I.e., Redhat Linux) distribution yet.
Compatibility and API issues with AWS:
- These problems will be solved because there will be demands for it to be solved. The fact that compatibility isn’t there won’t stop people from staying the course and building internally. There are 1000 companies in the space attempting to make cloud more or less usable. Any number of these companies will gladly jump on board to solve any specific problem. Will this make IT complex and convoluted, maybe, but that never stopped us before.
One or more major players dropping out (I.e., HP or IBM):
- I find this risk unlikely to occur, but it could. However, even if it does the only way it would have a serious impact is if it were to be two or more major players before OpenStack makes it to version 1.0.
Am I saying Public Cloud Players will be hurt?
One of the more interesting things about many of the arguments being made in the IT space is that we often make statements that suggest we see an “all or nothing” answer to major shifts in strategy or solution options. The fact is that the market for IT in general is only getting larger and the broader adoption of cloud in general is likely to accelerate that growth. It’s also true that the entire public cloud market is currently less than $4 billion, and the top six companies in the US spend more than that on infrastructure alone. This means that our favorite public cloud providers still have a giant opportunity for growth and this includes fairly large niche markets for specialty providers.
Go forth and Cloud
Regardless of what I’ve said, the theme of the arguments made by all three of the gentlemen blogs referenced here is that IT is not about the technology, or who’s providing it, but rather how using IT effectively can help accelerate the business. As long as you keep that theme in mind, spending a little more or less on your IT infrastructure won’t be the issue. The only issue will be whether you’ve helped make the business more competitive by leveraging all available options. So go forth and cloud, but take an umbrella.