All products 
Rocky Linux on a VPS: what it's for and what to check in a plan
Rocky Linux was founded by Gregory Kurtzer, one of the creators of CentOS, shortly after Red Hat announced that CentOS would move to the Stream model. The project is developed by the Rocky Enterprise Software Foundation, and its goal is stated plainly: the system should behave like RHEL, bugs included. That is why Rocky Linux is chosen where predictability matters: enterprise applications certified for RHEL, clusters, scientific and engineering computing, infrastructure that has been run by the same Ansible scripts for years.
Who Rocky Linux suits on a VPS
- Teams that run some of their servers on RHEL. On a rented VPS there's no need to pay for a Red Hat subscription, while packages, paths and systemd unit behaviour are the same, so playbooks and documentation carry over unchanged.
- Commercial software that requires "RHEL 8/9 or compatible". Vendors of databases, monitoring systems and backup tools often list Rocky Linux explicitly among supported systems.
- Kubernetes and container platforms. Nodes on the same system are easier to update in batches; plans for a cluster are on the VPS for Kubernetes page.
- Anyone moving off CentOS who wants the closest possible match to the old system. The migrate2rocky script handles in-place migration, but it is more reliable to set up a new server and move the data.
For a typical website under a panel, the difference from AlmaLinux is barely noticeable: both systems are free, compatible with RHEL and supported by cPanel, Plesk and ISPmanager. AlmaLinux turns up in host templates slightly more often and ships separate builds for older CPUs, while Rocky is closer to RHEL in the details. You can compare offers on the VPS with AlmaLinux page.
Versions and support periods
Each major version of Rocky Linux receives updates for about ten years, as RHEL does. Rocky Linux 8 is supported until 2029, Rocky Linux 9 until 2032, and Rocky Linux 10 until 2035. Within a branch, point releases come out (9.5, 9.6 and so on), and the server moves to them with an ordinary dnf upgrade. In-place major version upgrades are not officially supported: people usually move from 8 to 9 via a new server.
For a new VPS the sensible choice is Rocky Linux 9: almost every host has it, and the EPEL, Remi and third-party repositories are built for it. Go for version 10 if your software already supports it and you want the longest possible server lifetime. Keep the CPU requirement in mind: Rocky Linux 10 is built for the x86-64-v3 level and won't start on a virtual machine to which the host exposes a simplified CPU model. Before ordering, ask support which CPU the guest system sees, or check grep -c avx2 /proc/cpuinfo on a test server: zero means version 10 won't work.
A word about sources. Since 2023 Red Hat has published RHEL source code only for subscribers, and the clones obtain it by other means: from CentOS Stream, from the open UBI images and from cloud RHEL instances. Rocky Linux keeps the goal of full compatibility, while AlmaLinux has moved to compatibility at the application binary interface level. For a server with typical software this changes nothing, but if you are buying commercial software, check which clones exactly the vendor lists as supported.
Template or your own ISO
A ready template is convenient, but hosts build it their own way: some leave only a minimal set of packages, others disable SELinux or install their own kernel. If you need the system exactly like the company's other servers, with your disk layout, LVM and SELinux enabled, install it from the official image. This is possible on most KVM plans; the options are listed on the VPS with custom ISO page. For automated installation, a kickstart file comes in handy, the same one you use for physical machines.
Resources and first steps
A minimal Rocky Linux install takes a couple of gigabytes and runs on 1 GB of memory, but dnf uses a fair amount of memory when refreshing repository metadata, so add swap on the smallest plans. For an application with a database, plan on 2–4 GB of memory and a 40 GB disk; for a Kubernetes node, 2 cores and 4 GB or more.
- Run
dnf upgradestraight away and reboot if the kernel was updated. - Don't disable SELinux "just in case": most errors are solved by setting the correct labels with
restoreconandsemanage. - Check firewalld: in templates it is sometimes off, with SSH open to the whole internet.
- For a web administration interface there is Cockpit, which installs from the base repositories.