What Happened to OpenStack?

“Even then, if you’re using Kubernetes, you probably won’t succeed, because it isn’t in Google’s best interest to let anyone else actually compete with GKE.”<p>Google succeeds if it is anybody but AWS proprietary solutions. If they can groom a healthy ecosystem of open source and commercial solutions that target Kubernetes, then the tremendous advantage of AWS being a one stop shop for any service you can imagine starts to dwindle. As of now, amazon offers a solid compute environment and services galore, which is hard to compete with.<p>The author didn’t do much at all to tie what happened to OpenStack with Kubernetes. K8s is deployed at scale by all the cloud providers. Both Google and Microsoft solely run containers on it in their public cloud (while AWS still has their own orchestrator). That never happened at scale with OpenStack.<p>And regardless of how you feel about Google, Azure has a very strong vested interest in K8s success.

&gt; I am publishing this now in the hope that it can serve as a warning to everyone out there who is investing in Kubernetes. You’ll never be able to run it as effectively, at the same scale, as GKE does – unless you also invest in a holistic change to your organization.<p>This is a meaningless argument. I don&#x27;t have to run Kubernetes at the same scale as GKE to develop--I just run minikube, which runs very well on Linux hosts. When I get ready to deploy there is a pick of environments to host on because Kubernetes apps are largely portable.<p>OpenStack has never achieved this level of accessibility.

I worked a little bit with OpenStack about four years ago, and my impression was that it was very design by committee. Design by committee doesn&#x27;t work too well in software: <a href="https:&#x2F;&#x2F;sourcemaking.com&#x2F;antipatterns&#x2F;design-by-committee" rel="nofollow">https:&#x2F;&#x2F;sourcemaking.com&#x2F;antipatterns&#x2F;design-by-committee</a><p>I think a lot of the enterprise companies supporting OpenStack, like Mirantis (<a href="https:&#x2F;&#x2F;en.wikipedia.org&#x2F;wiki&#x2F;Mirantis" rel="nofollow">https:&#x2F;&#x2F;en.wikipedia.org&#x2F;wiki&#x2F;Mirantis</a>), realized this one way or another, got themselves acquired, and then used the new funding to pivot to Kubernetes or another open-source IaaS offering: <a href="https:&#x2F;&#x2F;www.mirantis.com&#x2F;" rel="nofollow">https:&#x2F;&#x2F;www.mirantis.com&#x2F;</a><p>Without any promise of enterprise support, there&#x27;s really no way for the large companies targeted by OpenStack to adopt it and make that adoption sticky. So that&#x27;s how it died.

My opinion: it grew to a gross, large set of python and apis that, when combined with multiple implementations, extensions, and company-specific customizations, made it an unmaintanable mess that was difficult to deploy and code around.<p>So, although the author compares it with the growing k8s project out there now, at least k8s more clearly stewarded, more developer oriented instead of only for ops (with code quality to match), and doesn&#x27;t feel as hamstrung by environments and dependencies (just try to run a little openstack setup on your laptop for development... very annoying for a project of such age with so many company&#x27;s hands in the pot).

My own impression.<p>If you develop on AWS you get a supported experience for a long LONG time (see simpleDB which I used and still works even though they don&#x27;t seem to market it). Same thing with old instance types. S3 etc etc.<p>With openstack at least a year or two ago - who can seriously stay on top of what is going on there. You could develop something 3-4 years ago and getting it going on the latest open stack = total pain. What exactly open stack was also muddy - lots of ifs&#x2F;buts&#x2F;this 5 year old code that ran on vendor X openstack doesn&#x27;t seem to run today on vendor Y.<p>Didn&#x27;t spend much time on open stack though - and I know the hype train was &#x2F; is huge - (AWS killer etc). My own sense - a lot of folks freaked out about AWS and all WANTED openstack to work so they had some big gun to blow up AWS with - but they didn&#x27;t seem to spend much time talking to actual customers &#x2F; developers, while AWS certainly did.

Openstack got good enough for people who wanted to run their own EC2 like internal cloud, right about the time that people stopped wanting to run VMs.

I&#x27;ve never seen an openstack implementation that wasn&#x27;t horrible. The reason more people use k8s is it&#x27;s actually less useful than openstack, which makes it more opinionated, which makes its implementations more uniform. Plus, most people deploy it either on top of an openstack or other cloud platform.<p>You should not build your own cloud platform; that much is obvious. It&#x27;s less obvious that you should not build your own k8s, because it seems simpler and more useful.

I think the battle for running virtual machines in the enterprise was already won by vmware and so OpenStack became a niche for service providers wanting to do Nfvi workloads and had resources to afford dedicated teams to running OpenStack.<p>Kubernetes on the other hand capitalized on the need for running and orchestrating containers. Kubernetes also got a few things that right such as well documented and prescriptive set of tools one could use to get dev and production cluster up.<p>On a separate note, having worked on OpenStack I can also attest that the code was gross, not so much in kubernetes.

&gt;Enterprise customers. That’s a nice way of saying well-known brands who spend a lot of money on each other every year<p>i&#x27;m definitely going to be stealing that line

When I used to use OpenStack, it frustrated me because the software seemed badly designed, unreliable, poorly documented, and hard to use.<p>Perhaps having hundreds or thousands of contributors is more of a problem than a solution. Or maybe it just needed better technical oversight.

We are working on a project where we plan to manage our new HPC system with OpenStack. This will replace 3 legacy HPC systems that were manged with propriatory management systems. Due to shifting requirements by our customers (scientist) we decided to move to a cloud framework where probably still the majority of resources is dedidcated to a batch scheduling system but it would allow us to also provide more cloud like services (jupyerhub, rstudio, databases, etc) It&#x27;s quite an ambitious project and we (4 engineers) basically spent the last year understanding the in and outs of OpenStack. We also went full in with integrating all kinds of datacenter components into OS (NetApp, SDN, DNS, etc)<p>Some lessons learned so far:<p><pre><code> - OpenStack is very complex - It&#x27;s less of a product and more a framework and you need a dedicated engineering team with cross cutting skillset - You definately need a dev&#x2F;staging environment to test upgrades and customizations - Some of the reference implementations of OS servies (SDN) are fine for small deployments but if can replace them with dedicated hardware&#x2F;appliances you should do that.</code></pre>

&gt;When you’re looking at other cloud products, think about similar conflicts of interest that might be affecting your favorite spokespersons today… (I’m looking at you, kubernetes)<p>See also: Banks investing in cryptocurrency R&amp;D

Funny enough I&#x27;m deploying OpenStack right now for us as an internal playground. It&#x27;s decent - but hell, the learning curve is nasty and the documentation is incomplete. Many things I could only get working after asking on IRC and waiting hours for a reply.<p>But still, it&#x27;s better than having to manage KVM by hand and cheaper than buying VMware.

OpenStack was an extremely ambitious project, which requires so much cooperation and interoperability between so many companies that I believe it was doomed from the start.<p>In 2015 my company purchased &quot;Flexpod&quot; which is a solution that&#x27;s certified by VMware, Cisco, and NetApp to work together. The result is nothing but a bunch of back and forth finger pointing with support, and even a critical vulnerability will take 6+ months to get patched and certified between all the different vendors.<p>I personally like the Ansible approach where each Storage&#x2F;Computer&#x2F;Network vendor provides APIs for management of their devices, and Ansible is just the glue between them.<p>TL;DR getting major tech vendors to play nice together is hard.

Luckily there is also OpenStack running on top of Docker... <a href="https:&#x2F;&#x2F;github.com&#x2F;openstack&#x2F;kolla" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;openstack&#x2F;kolla</a>

NephoScale was an open stack vendor I followed before they vanished.<p>Seems like it never really got easy enough for average developers to find it accessible.

I&#x27;ve been using AWS for years, and had never heard of OpenStack until this year.<p>The only reason I&#x27;m aware of it, is because I&#x27;m studying for a degree part time - and OpenStack is taught in one of the modules. It&#x27;s a shame really that they only mention AWS and advise against using it in case you accidentally spend money.

As an outsider, the concept of openstack was always really appealing but it felt like RackSpace’s answer to AWS. Never really learned more about it than what was on their website but that was always the impression I got.

I think the question is not &quot;What happened to OpenStack&quot; but rather, &quot;Is OpenStack still total garbage?&quot;<p>I&#x27;ve been at two companies that attempted to go down the OpenStack route. One wanted to start a cloud offering to their clients and hemorrhaged tons of money trying to just keep OpenStack stable. We couldn&#x27;t even run our basic Logstash offerings on our OpenStack cluster without them having bizarre performance issues.<p>We had a really good manager too who had accounts on every other provider (Rackspace, RedHat, Canonical .. all the big ones) and time and time again he was like, &quot;What is this? How are they doing this..&quot; and we just figured they used a ton of specialized proprietary plugins they just weren&#x27;t open sourcing or a ton of special patch sets.<p>Second shop had tried moving onto an OpenStack cluster to save on AWS prices. It could never run anything reliably and they scrapped the entire project and re-purposed all the servers for DC&#x2F;OS, which was super nice and reliable and every team migrate hundreds of services onto.

This way of writing (like a journalist) is so annoying. I prefer much the scientific style of writing where the main idea is given in the first sentence, and suspense is avoided as much as possible.