One of the most frustrating aspects with companies, especially smaller ones (as if often the case in the open source ecosystem) is that they do not explain the philosophy behind their product. And in some cases, I’m not sure they have a well defined strategy or communicate that clearly if it changes.
For Univention my perception is changes of scope and direction, including contractions. And while expansion in any direction is usually transparent to customers, contractions can be much more difficult to manage by clients and result in critical loss of trust.
Just to highlight, here is my personal perspective:
Univention seems to have started off with little more than the idea, that you can offer big savings and make a quick buck for companies that would otherwise have to buy Windows NT servers by using Samba and invest a bit into making that look better and work more easily.
That’s legitimate and fine, especially when you offer the result back as open source, so please let me assure you that this isn’t my gripe, more how I try to understand the somewhat erratic changes in the philosophy of your company, which I guess could ill afford to [re-]invent itself without the help of venture millions or a M$-style practical monopoly.
Your founders then seem to have looked for growth beyond that initial idea, but what exactly Univention wanted or wants to be, has actually become less and less clear for the 10 years or so that I’ve been observing you.
I noticed signs of significant functional expansion around UCS 4 with what seemed like a a strategic move towards becoming a provider of IAM integrated turnkey solutions that pointed [application stack] upwards, towards groupware applications like Open/NextCloud, but you were also offering [application stack] downwards instrumentation around IAM integrated VMs and containers, which to me looked a lot like HCI or at least hypervisor management, perhaps even orchestration: not quite a Nutanix or vShere drop-in replacement yet (nor Parallels/OpenVZ), but perhaps at least at the Proxmox level of management integration.
That integration is huge value, especially when unlike M$ it comes without having to sell your kidneys, so that’s how I came to invest my time on your product and tried evaluate if it could become a product for my employer, then a company with 100k employees and a way bigger IT customer base. Its CEO even came to be an EU commisioner until recently and while I was able to impress my hyperscaler gripes on him before he left for the EC, he (and we) still lost to the clouds on all fronts…
…and Unvention never left my small lab (originally mostly based on CentOS7/oVirt, also OpenVZ, now based on Proxmox), because its scope and productivity advantage never really scaled beyond what it included without major integration effort.
The single management pane for the entire stack is part of how clouds killed on-premise and with a giant budget and re-engineering effort even Microsoft escaped with their Office into 365 and Azure with AI to suck the world’s biological brains on every keystroke.
That is a tough place to compete for a small German company, but (what I perceive as) a functional shrink almost back to your AD (or Novell directory services) replacement starting point isn’t going to cut it, even if you now try to focus on deployment scale-out up to to government levels and bet on regulatory protection by them.
Yes, identity management is at the very beginning and core of a full stack, but at least Herr Ganten still remembers Novell’s fate and that directory services alone aren’t enough.
That new modesty may be hard earned realism, but even with Trump poisoning the cloud it seems that Microsoft and Musk lobbying might just get replaced by having their people in direct political power, eliminating the regulatory protection you seem to have chosen to build your future on: the new sovereign may not be the people any more.
And that is why the vision of a full stack ecosystem which starts with a single box and grows organically via “we the people” to the size of federated nations needs to be in your focus, and it needs to work in practice.
It obviously can’t and shouldn’t be a single vendor solution, but it has to have the single vendor look and feel. That only works if everyone goes quite a bit beyond their home turf, but in a good way.
If I understand you correctly, in your vision you would like to install Proxmox and UCS on the same Debian based physical machine. This is indeed not supported by UCS (and I assume even not by Proxmox), and even if it would be supported I wouldn’t recommend it. Splitting Infrastructure (like Hypervisors) from Applications (like UCS and Nextcloud) is best practice for a good reason - you can more easy exchange one or the other (for example exchange the Hypervisor like now intended by many who use VMware).
This paragraph to me shows everything that is wrong with the current Univention approach: of course, you’re technically right, but IMHO you’re wrong, functionally and philosophically.
You are putting yourself into a functional niche, concentrating more on what you don’t do, because it’s another vendor’s job, instead of what what you can help to achieve via what could function as a seamless integration.
And that what gets you into the non-Amazon, Meta, Azure or Alphabet graveyard Martin Andree talks about in his Univention Summit 2025 video.
What I want is what I believe many others want, too, because I consider mayself quite ordinary: the deep-enough stack, fully integrated under a single management pane in a turn-key package. I want simple deployment and easy organic scale. I can start with a single box and add standby and cluster as befits the environment. I want the flexibililty to move from boxes I own and house to clouds scattered across the countryside or even continents. I can also cut and fork, because those are natural processes in families and even with EU nations.
Whether I deploy on physical boxes from e.g. a USB stick with a Proxmox or Unvention boot medium, should be following a shared set of use-case based (tested) recommendations from both. And once those boxes are bootstrapped, adding UCS for the management of those boxes and the applications on top should be just as easy as adding a seed VM/container/app and having it go to work with a few quick steps, again with both sides recommending a simple procedure to achieve the integration.
Currently both sides leave all the integration effort to the end-user or potentially some 3rd party. Yes, supporting all types of flexibility there is a hard need, but no, you shouldn’t be left with a stack of manuals on your own by default. And most importantly, unless you can rest assured that a rich and complex product, which will require quite an investment to learn and appreciate at the application level, will continue to work and operate with the underlying stack staying abreast of security threats, but also evolving with fundamental changes in the state of the art, customers just won’t cross the threshold and commit potentially man years of effort into that eco-system they see as an integrated stack.
The value isn’t created by the infrastructure, nor by the identitiy mangement, but at the application level. Unfortunately deficits at those lower levels, don’t just destroy application value, their negative impact can be much bigger as the Mirage botnet demonstrated.
Leute, ihr müßt eure Ambitionen deutlich höher ansetzen.
Grundsätzlich ist jedes bißchen Zeit, was ihr darauf verschwendet, den Leuten zu erklären, daß sie wegen falscher Abstraktionen zu fehlerhaften Aufbauten neigen, teuer verschwendet. Statt dessen solltest ihr eure Energie darauf verwenden, die Integration mit komplementären Partnern zu vereinfachen: Proxmox ist global gesehen doch nun wirklich um die Ecke und seit Jahren scheint ihr mehr darauf bedacht, euch nicht auf die Füße zu treten, als einander besser zu integrieren.
Bin ich wirklich der einzige, der mal eben aus ein paar NUCs oder Laptops einen voll integrierten NextCloud/UCS-school/org Server[-Cluster] mit so viel Redundanz und Skalierung haben will, wie ich extra Maschinen dazustellen möchte?
Die aktuelle politische Situation könnte eigentlich kaum besser für euch sein. Wie gefährlich die Verlockungen der Wolke für jede Form von Souveränität ist, war nie deutlicher. Und wenn ihr diese Gelegenheit nicht für eine philosophische Neuausrichtung nutzt, dann werdet ihr sehr bald auf dem von Herrn Andree so anschaulich beschriebenen Friedhof landen.
Und weil das Thema KI euch so am Herzen liegt, will ich dazu noch ein paar Takte sagen, die ich mal aus dem Hype um IoT heraus entwickelt habe:
Niemand will intelligente Assistenten, nicht einmal biologische umsonst, statt künstlicher gegen teuer Geld, wenn diese nicht garantiert loyal und langfristig verfügbar sind. Kaum jemand wird eine Putzfrau engagieren, wenn er sie nur zwei Stunden bekommt und garantiert jedes mal eine andere ist: Bis man der den Job erklärt hat, ist man heiser und die Dame wieder weg.
Ein Butler wie Carson in Downton Abbey erfordert nicht nur eine grandiose Ausbildung, sondern wird vor allem dadurch wertvoll, daß beide Seiten zu gleichen Teilen in den Aufbau einer gemeinsamen Wissensbasis oder gar Kultur investieren und er eine Familie über ein ganzes Leben begleitet. Außerdem darf er natürlich nicht jeden Schritt an seinen Ausbilder zu berichten oder gar Anweisungen von dort entgegennehmen. Und er übernimmt das Haus, egal wie groß oder klein, skaliert die anderen Dienstboten transparent über Delegation, koordiniert mit den Butlern anderer Familien bis zu dem Punkt, wo eben Andeutungen reichen, um physisch große Dinge zu bewegen.
Hier steht den Internetgiganten bei der echten Wertschöpfung ihr Geschäftsmodell entgegen, selbst die Generation-Z wird irgendwann überlegen, ob sie wirklich die Wiegen ihrer Säuglinge von Tesla-Robotern schaukeln lassen wollen: convenience isn’t everything.
Aber bis die diese Hürde überwunden haben, muß der wirklich souveräne Stack, der bei Handy in der Hosentasche anfängt und dann organisch mit den sozialen Banden von der Familie bis zum Staat skaliert, auch technisch funktionieren: Da seid ihr gefordert.
Ihr habt meine E-Mail über das Forum, wenn ihr reden wollt, tretet in Kontakt.
which I forgot about 
