Tuesday, July 16, 2013

The trouble with OpenStack

Depending upon which analyst or flat earther you listen to, then OpenStack rules. I don't buy this for a second and instead take the view it's most likely to be a dead duck. Why?

Well, from its very beginning, when I put OpenStack on stage at OSCON Cloud Summit in 2010, I warned several of the executives behind the project on the dangers of Amazon's ecosystem. Being an Amazon clone and co-opting was priority number one in my view. Alas, they wanted to talk about differentiation and not being beholden to Amazon. This was their 'strategy' in a commodity market and they ignored that if you grow your ecosystem to be the larger then you actually control the API. Since that time, I've heard this story of differentiation repeated over and over again. It felt as though 'anything but Amazon' ruled their thinking rather than sound strategic sense.

Now, I need to be clear here. This wasn't everyone. There was one particular individual (who I previously worked with) behind OpenStack who really understood the importance of co-opting and how to play a good game - Rick Clark, Unfortunately that battle didn't seem to be won. Personally, I take the view that if Rick had been given entirely free reign in the project then things would have been very different. 

Anyway, that isn't what happened and so today we find that OpenStack has a significant developer and vendor ecosystem but is categorised by a collective prisoner dilemma. Forget fidelity with Amazon there isn't even fidelity between OpenStack distributions and no well established mechanism of doing this (though first steps have been made). The solution to this problem and the use of trademarks has been well known for many, many years but to compound it the focus of OpenStack also seems to have shifted to private rather than public as though avoiding the battle with Amazon has become a goal?

There are a some voices of reason in the mix, people like Randy Bias has a relentless focus on building AWS compatible OpenStack clouds (something I strongly agree with being an advisor to his company) but overall in my opinion the potential of OpenStack has been squandered by a heady mix of marketing and poor strategic play. Had only the wisdom of Mark Shuttleworth been listened to or potential customers like NetFlix's Adrian Cockroft - "please try to build AWS clones that scale".

Don't get me wrong, I'm all in favour of an open source reference model as the basis of a cloud market - I keynoted about the necessity of this at all layers of the stack at OSCON in 2007. Nothing has changed my mind since then but OpenStack IMHO failed to follow the path of pragmatism and give users what they've been screaming for - a market of public AWS clones. Even the recent survey of OpenNebula users reinforces the point. 

Does it mean OpenStack is out for the count. Well, not quite - as I mentioned last year it's possible if the players behind OpenStack flood billions into the system and create a massive competitive market around OpenStack by end of 2013 then they can still compete with this whole differentiation play. I see the hope of that happening as somewhere between Bob and no and Bob has left town. Time is running out, hope is fading fast.

So, is that it for OpenStack in my opinion? In my view, yes. Short of a miracle such as OpenStack announcing it'll become an AWS clone or some new entrant coming into the market, grabbing OpenStack and building the world's largest public AWS clone by flooding billions into its creation then I can't see it ever reaching its potential. In fact, I see things disintegrating as VC's start to ask "where's our future money?" 

Do miracles happen? Unlikely, you'd need a company with huge amounts of capital (many many billions) and vision, a great strategist and practitioner who understood the importance of ecosystems and you'd have to build a rock solid engineering team and large scale operations in lightning fast time with a brutal commodity focus. That seems like wishful thinking, the song "livin' on a prayer" comes to mind.

So barring miracles or somehow a massive market forming in the next six months then I hold to the view OpenStack is a dead duck. I don't see that changing. It shouldn't have been this way, it should have from the very start been an AWS clone.

-- Update 16th June 2013

There are a couple of other scenarios where OpenStack might succeed beyond forming a massive market by end 2013 (unlikely), a u-turn on the whole differentiation play (unlikely) or some other miracle (unlikely). This includes co-opt GCE (Google Compute Engine) or magic lobbyists somehow persuading Government's to ban Amazon or adopt some other API set as the standard. The former is possible and raises its own set of questions, the latter assumes some rather strange behaviour by Government's and seems pretty unlikely.