Managed Kubernetes · your hardware
Kubernetes only works if someone installs it properly, upgrades it without taking your product down, patches the machines underneath, and proves the backups actually restore. That is a full-time specialist job.
KubeNest is that job, as a subscription. One fixed monthly price, on hardware you already own, for less than the cost of hiring one platform engineer.
Delivered with Concinnity Solutions. Runs on your infrastructure — we never hold your kubeconfig.
kubenest install --bundle 1.4 \ --hosts 10.0.1.11,10.0.1.12,\ 10.0.1.13k3s v1.31.4 readytraefik + gateway readycert-manager readyopenebs local-pv readyvelero readyupgrade controller readyThat is the whole install. Everything your applications need to run — networking, certificates, storage, backup — set up together and tested as one thing.
What is it?
We install a complete, tested platform on machines you already own — in one command, usually in an afternoon. Then we keep it running: upgrades, security patches, backups and monitoring, with a report every month showing what we did.
What does it fix?
Kubernetes on its own doesn't run anything. You also need networking, certificates, storage and backup — chosen, wired together, and kept compatible with each other forever. That work usually lands on one or two people, who then can't take a holiday.
Why not the alternatives?
Amazon and Google run the small part in the middle and leave the rest to you, on their hardware. Rancher gives you a dashboard, not a maintained platform. Hiring costs more and walks out of the door when the person does. The long answer is here.
How do I get it?
There's a free edition you can download and install today, without talking to anyone. When you want it run for you, it's thirty minutes on a call, a fixed monthly price in writing, and we take over. No procurement marathon, no twelve-month commitment before you've seen it work.
Nobody budgets for "keeping the platform alive", so the cost shows up somewhere else. In three places, usually:
In salary.
The median US platform engineer earns $131,607 in base pay1 — before benefits, payroll taxes and recruitment fees, and before the six months it takes them to become genuinely productive. Most companies need more than one, to cover holidays and illness.
In delay.
Upgrades get postponed because nobody is confident they're safe. You fall several versions behind, and then the upgrade genuinely is dangerous — so it gets postponed again.
In risk.
Backups get configured once and never tested. You find out whether they work on the worst day of the year, in front of your customers.
We replace all three with a fixed monthly price and a report you can hand to your board.
01 — What we install
You don't choose components and hope they work together. We ship a set we have already tested as a unit, and we upgrade it as a unit. You run Platform 1.4 — one number you can put in a compliance document, not eight moving parts installed on different Tuesdays.
If something in that set has a security problem, patching it is our job.
| Component | Pinned | In plain English |
|---|---|---|
| k3s | v1.31.4 | Kubernetes itself |
| Traefik + Gateway API | v3.3 | Lets the internet reach your apps |
| cert-manager | v1.17 | Keeps your HTTPS certificates valid |
| OpenEBS Local PV | v4.2 | Fast disk for databases and files |
| Velero | v1.15 | Backups, and proof they restore |
| system-upgrade-controller | v0.14 | Upgrades machines without downtime |
| kured | v1.16 | Reboots servers one at a time, safely |
| KubeNest agent | v1.4 | Tells us when something needs attention |
Example bundle. Exact versions are published per release with a test report.
Free edition
You can download KubeNest and put it on your own servers today, without speaking to anyone and without a purchase order. It's the same installer we use — the same tested set of components, the same single command.
The free edition proves the install works. The subscription is what stops you having to think about it again.
02 — What we do after it's running
Every team we spoke to said the same thing, and none of them said "deployment". The problem was the year afterwards: upgrades, patches, and finding out too late that the backup had never been restored.
Before anything moves, we scan what you're actually running for anything the new version will remove — and we stop the upgrade if we find it. An upgrade that succeeds and takes your product offline has not helped anybody.
A backup nobody has restored is not a backup, it's a hope. On a schedule, we restore yours somewhere safe, check it came back correctly, throw it away, and record a pass or a fail with a date on it.
Kubernetes upgrades get all the attention. Unpatched servers are what actually gets you breached. We patch the operating system and restart machines one at a time, so nothing goes down while we do it.
Our agent calls out to us — it opens no ports and needs no firewall changes. Certificates about to expire, disks filling up, machines that have dropped out, patches missed, backups not tested lately. If a cluster goes quiet, that is itself an alert.
03 — Why not just…
Some of these are good products, and for some companies they're the right answer. Here's where each one stops.
Their managed Kubernetes runs the coordination layer in the middle — a genuinely hard part, and a small one. Everything your applications actually touch is still yours to choose, connect and maintain, and it all runs on their hardware in their region. If that's fine for you, use them. If data residency, cost at scale, or servers you already own say otherwise, you still need everything we do.
Rancher is a good control panel, and looking at your clusters was never the hard part. It doesn't decide which networking, storage, certificate or backup tools you run, and it doesn't promise they'll survive the next upgrade together. That assembly and testing is the actual work. Rancher's automated maintenance features are a paid tier anyway.
Well-built products. They also require their own operating system on your machines, so if you run Ubuntu today that's a migration project before you get any benefit at all. We install onto the servers you already have, as they are.
You can, and past a certain size you should. It costs more than this, takes months to recruit and months more to become useful, and the knowledge leaves the building when they do. We're a team rather than a person, under a contract, with the work written down and handed over.
Plenty of teams do, and it works — right up until a certificate expires while the person who set it up is on holiday, or an upgrade needs a maintenance window nobody wants to own. If you're reading this page, you probably already know which one of those is coming.
04 — No lock-in
Everything we install is standard open source that thousands of companies already run. There is no KubeNest-only technology holding your systems together.
If you stop paying us, nothing switches off. Your clusters keep running exactly as they are, and any competent engineer — yours or someone else's — can take over. You're buying the assembly, the testing and the upkeep. Not a trap.
KubeNest itself is commercial software. Everything it installs is not.
Run KubeNest for your clients and see all of them in one place. You keep the relationship and the margin. We keep the platform tested, the upgrades safe and the restores proven — and give you the monthly evidence to put in front of your clients.
Data that can't leave the country, costs that stop making sense at scale, or hardware you've already bought — whatever keeps you off the big providers, you still need someone to run the platform layer. That's us.
Thirty minutes on a call: what you're running now, what's causing you trouble, and whether any of this is worth your time. No slides, and no obligation to go further.