Let a thousand public clouds bloom

As the world is admiring how well Google Cloud has caught up and how they are offering an immense competition to AWS, I cannot but lament the golden opportunity that was wasted by OpenStack community.

As you know, I am a public cloud proponent.  But, I am not a fan of centralized power between just two players in the industry – it is bad for customers and innovation. Unfortunately, we are headed down that path for foreseeable future. We have Amazon, Google and mostly also-rans lagging far behind. (Expect next IAAS MQ to have these two guys in upper quadrant and rest in the lower middle).

Where has OpenStack gone wrong? It tried to be an Uber solution. It failed to make choices. It let time pass and hoped inaction will somehow fix things. It lacked decisiveness. It spent more time arguing and preventing good people from furthering their case than making progress. It refused to listen to good outside people.  I get it is an open source project. I get that there are too many cooks in the kitchen. For all the millions spent by companies working on it, if they stepped back and considered the market dynamics, they perhaps could have made different choices. All I see is every player mapping how OpenStack be a supplement to their existing business. Their resources, processes and values are optimized for status quo.

There are many hosting companies. These guys were the innovators in the 90s. They dropped out of schools to start these hosting companies (and I am not just talking about RAX, there are many others that fit this profile). They moved to virtualization when VMware innovated in the space. They were rewarded well for that choice.

After virtualization, Amazon came around with AWS. It was new innovation. Amazon is a closed and greedy company – they would not share neither their technology nor let these hosting companies play in the ecosystem well. VMware fell into classic Innovator’s dilemma and is digging a deeper hole that it cannot get out of. VMware is not the solution these hosting companies need.

This was an opening OpenStack had. When AWS gained ground, it gained ground with developers.  AWS smartly went beyond developers and started appealing to production workloads. It was either you do AWS or you refuse to go public cloud. In many companies, VP of App development often wins against VP of Operations in arguments. So, whether they liked it or not, VP of Ops had allowed use of AWS. Amazon is winning this war. Google will take its share as well.

What could OpenStack do? Instead of scratching the itch of its backers to make enterprise sales, they should have focused on the hosting companies. They should have made OpenStack the default choice for the operators. They should have made it stupidly simple to co-exist and interoperate with AWS. They should have embraced AWS for dev usage with open arms. Instead, they viewed AWS as an enemy before they were the enemy. They could have removed the friction of moving to public cloud. Moving production workloads to AWS is still a disruptive process for many companies. Migrating to public cloud without changing out your vendors and giving up owned infrastructure is less friction than outright moving to AWS/GCE ( contracts, getting used to support model, relationships are just few examples of friction ).

I believe OpenStack still has a small choice to become an attractive choice for operators. But, first they need to stop talking about private cloud for end customers. Private cloud is a fool’s errand, they are not only fooling themselves, they are screwing their customers too, because 100% of private cloud deployments will fail.  They have same problems of owned infrastructure that IT wastes millions of dollars on.

OpenStack needs to be laser focused on cloud operators and let many public clouds come to market that can play nicely with each other and embrace AWS and GCE as part of the ecosystem. You may say, but some operators already use OpenStack – the question is whether its possible or not, its whether they are focused on it or not.

Oh, and for not a minute, do not think OpenStack can move into ill-defined PAAS space or a container is going to be their savior. Case in point, we are currently witnessing the slow painful death of Solum.

You know, I keep thinking perhaps, it’s not OpenStack, but CloudStack that may enable this, but I am not too familiar with them. Perhaps, should take a closer look.

Industry needs robust set of competitors – a two vendor market is not a good future for IAAS. We need more.

Let a thousand public clouds bloom.

Advertisements

3 thoughts on “Let a thousand public clouds bloom

  1. Please wake up from 2010. You’re providing a simplistic view as if public cloud is error free and private cloud is useless. While the promise of anybody can be a public cloud provider is attractive and possible, no sane customer would use a public cloud run from a garage. OpenStack is correct in targeting enterprise and private cloud is a big part of it.

    Please show me one reasonably sized customer who wants to move from one cloud to another. If someone is on Foo, they don’t move from it. Unless Foo dies. So Embracing AWS or GCE compatibility helps someone’s insecurities, but doesn’t help anyone really.

    I agree that industry needs more, but not just public. And the pie is bigger. I won’t call anyone a winner yet.

    • Boss, in this comment above we have here someone that has done their homework…
      In related news our K- cloud special forces goats, deeply infiltrated behind enemy lines confirm that:
      – Each and every telco (i.e. “Service provider”) investment in Dope’n’Stack that we know of is fundamentally & irreversibly fucked up
      – The very few “pure public cloud” providers using OpenStack are not doing so well
      – The table stakes to be a credible player in that space run well into Billions, and few have the guts & appetite.

      What you _do_ (and IMO _ will_ have) is many small to “moderately big” intallments that will act as niche providers for a mix of internal / external parties. Their basic role is to act as a sandbox for safe learning & assimilating this radically new infrastructure model (idiosynchrasies btw AWS / GCE / etc notwithstanding). If there is _ actually_ a business case they sustain is a (far) second issue.

      Oh, and since it’ll take quite a while for this to play out it’s far to early to discuss what will come after & dismiss it as a “transition phase”. Cause at the end of the day, all is “transition” – the only eternal constant is Ardbeg.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s