This era is really about managing multicluster, multidistro Kubernetes across a spectrum of different environments at the same time.
When we launched our State of Production Kubernetes research project three years ago, our goal was to track and validate a major ongoing shift in the cloud native and Kubernetes space.
We observed growing complexity across many different aspects of enterprise Kubernetes. Organizations were running Kubernetes in more environments, with more clusters and more software elements deployed in each stack and using more Kubernetes distributions across their estate.
In our latest edition of the research, announced at KubeCon in Chicago, we found that 83% of organizations use more than one Kubernetes distribution today. Fifty-nine percent use between two and five; 9% use more than 10 different distributions!
This research paints a picture of rapid growth and vibrant experimentation. It also describes a new era where platform engineering teams are serving diverse stakeholders — managing production clusters across a whole spectrum of different environments at the same time.
Your Distro Isn’t Special. It’s One Piece of the Puzzle
Kubernetes distributions (or “distros”) come in many different varieties. They might:
- Be tied to cloud services, like EKS Distro (EKS-D) used by Amazon.
- Form an integral part of ecosystems of platform services that developers use, such as Red Hat OpenShift.
- Target very specific use cases — edge is a good example — with lightweight distros such as MicroK8s or K3s or our own PXK-E, purposely built for small factor appliances.
- Be built for security, with reduced attack surfaces, FIPS (Federal Information Processing Standards) cryptography and immutability.
- Optimize for simplicity, with a fully packaged stack, from OS, Kubernetes distro, to add-ons and applications.
- Embrace openness or come with strong dependencies or “opinions.”
Whatever the driver, the reality is that the more organizations expand their Kubernetes footprint, the more they will have to treat the distribution as just another piece of the puzzle — not as the puzzle itself.
Many individual maintainers, communities and vendors have put in hard work to innovate and build hundreds of Kubernetes distributions out there. That work has value. But no single distribution is right for every use case.
Organizations need choice and the tools to operate more than one distribution simultaneously — and the same applies to operating systems and all the many software integrations that make up a complete production cluster. This era is really about managing multicluster, multidistro, multienvironment Kubernetes at scale.
Only One ‘Opinion’ Matters: Yours
The multi-everything reality is why we built Palette the way we did. No platform should be highly opinionated and lock you in. We set out to solve real-life Kubernetes problems for our customers, irrespective of the “flavor” or vendor and origin of their Kubernetes stacks. And we found that our customers really needed choice, open standards and support for their existing environments.
During KubeCon Europe 2021, Ben Beeman, of GE HealthCare, referred to a challenge we helped the company to solve: He talked about a “solution to help them with all the cloud native chaos and multiple layers of software” (cloud native open source and commercial integrations) that they needed to manage the full-stack life cycle, providing not only flexibility in choice of integrations, but also consistency. It wasn’t about the distribution at all. It was about the bigger picture.
What Does Your Future Hold?
While “lock in” and “choice” might seem like overused terms, they are important. They point to a fundamental need: to futureproof investments without having to design an “exit strategy.” And this is especially important in today’s market, buffeted by constant change.
The reality of today’s IT landscape is this: Promising open source software projects fizzle out, as we saw with k3OS.
Vendors pivot, go bust or get acquired — most recently, D2iQ, formerly Mesosphere, laid off its staff, and the future of its Konvoy distribution and DKP platform is in the air.
Even giants are not always safe. Just recently the Broadcom acquisition closed, and thousands of VMware employees were laid off, with VMware partners reporting that unclosed sales and renewals are now in limbo; analysts at Forrester advised VMware customers to “brace for impact.”
If you’ve committed to a distribution, platform or service from a vendor like this, how can you mitigate the risk and move forward confidently?
The Answer: Prioritize Openness and Extensibility
For those of you facing uncertainty with your previous choices and want to protect your existing investments and extend them to the future, we are here to help.
We recently announced an offer for organizations that invested in D2iQ’s DKP and the Konvoy Kubernetes distribution, supporting their environment end to end with Palette and making the transition as smooth as possible, including providing accelerated training and a great cost incentive.
Earlier this year, we announced support for virtual machines with enough functionality packed into Palette to support production workloads. Our own distributions (PXK and PXK-E) are 100% hardened, curated, but also 100% Cloud Native Computing Foundation (CNCF) conformant. If you decide to leave Spectro Cloud, you can use them with any platform.
Last but not least, our entire platform is built on top of mature open source technologies such as Cluster API and its pluggable providers for cluster life-cycle management, KubeVirt for VM management, Helm and GitOps for application life-cycle management. You are getting a turnkey enterprise-grade platform but are not locked into it.
We remain committed to helping organizations already using any flavor of Kubernetes across any environment, workload and stack, acknowledging the need for choice to innovate and future proof existing investments. Why not put us to the test?