Interesting facts from someone who has dedicated much of their life to the administration and design of large systems.

The most widely used Linux distribution on servers is Ubuntu, because it has been able to move into the cloud and most providers offer it.

Ubuntu is a very good distribution for system administrators. It is essentially Debian with certified support for servers and a defined life cycle. Unlike Debian, we know when the LTS versions of Ubuntu will be released and can plan accordingly.

Many Linux distributions are less secure than Windows 11 by default, as they come without a firewall enabled by default, without AppArmor or SELinux, secure boot, etc. Always use distributions that come hardened by default, such as Ubuntu, Fedora, RHEL, Debian, or openSUSE.

If you want something that doesn't break, use Fedora Silverblue.

AIX is simply great; it's the operating system every sysadmin dreams of. Everything is perfectly documented through error codes. Its LVM was advanced for its time, with software RAID, snapshots, quorum, and distributed metadata, etc., and together with OpenBSD, it is one of the most secure systems in existence. Unfortunately, it only runs on IBM Power machines.

Solaris, another great system, especially with the arrival of ZFS, was a shame when it was acquired by Oracle. Solaris was a cutting-edge OS in every way.

With FreeBSD, I have hosted thousands of web servers and petabytes of information with ZFS. What can I say about this great, pure, hard UNIX? I will always carry it in my heart.

And finally, System Z and its z/OS (among others), unbreakable, can run at 99% capacity continuously and you won't notice any delay. You can run applications from 40 years ago; the backward compatibility is simply incredible. The security of OpenBSD next to z/OS is a joke. And like all IBM mainframe products, the documentation is simply spectacular.

As you can see, I have a fondness for mainframes. In this sense, Linux was a big step backward for large system administrators, as the documentation is often terrible, errors are continuous and often difficult to fix, especially in distributed systems. In exchange, the costs are lower, but the headaches are greater 😂 .