Red Hat Enterprise Linux is widely thought of as the first choice in operating systems for important servers, but it may not be the right choice for all applications.

The decision to use any modern edition of that operating system, generally spoken as RHEL with a silent H, is usually based on a need for component stability, paid technical support, and long-term version support, said Red Hat’s Ron Pacheco, director of global product management.

Customers have other options for data center operating systems. RHEL wouldn’t always be appropriate for edge devices, functions-as-a-service, and highly specialized applications, Pacheco noted.

SEE: Linux distribution comparison chart (Tech Pro Research)

Mitch Pirtle, an open-source advocate known for co-creating the Joomla content management system, pointed out that RHEL’s open-source cousin CentOS is the most viable alternative to the commercial version. His top advice for sysadmins: “First, define the needs of an enterprise Linux–are you looking for vendor support contracts, hardware support, or something else? I know that RHEL used to have device drivers that couldn’t be found anywhere else,” Pirtle said. “On the RHEL side I’ve only seen a few shops that are ‘RHEL shops’ that aren’t actually just running CentOS everywhere with RHEL only on core systems. I know of a few others that actually are all running Ubuntu, but their ‘standard’ server distro is RHEL and everyone just fills out an exception form.”

Canonical’s Ubuntu Server has good support, and Debian is also a strong option, Pirtle said. Hyperscale public clouds, Ubuntu-centric commercial application, systems with unique hardware requirements, systems with custom kernels such as for machine learning/data science, and specialized deployment requirements such as CoreOS itself are all additional examples where RHEL shouldn’t be the top choice, Pirtle explained.

Alpine (for security), ClearOS (for smaller enterprises), and Scientific Linux are still other distributions to consider.

Pacheco noted that the latest mainstream version of RHEL is 7.5, released in April 2018, which has functions such as virtual data optimizer based on the July 2017 acquisition of Permabit. That performs on-the-fly deduplication in memory before data is written to disk–a welcome addition for system administrations because this typically requires aftermarket storage management software. Also new in RHEL 7.5 is network-bound disk encryption, which uses software keys to unlock many virtualized instances simultaneously and which only applied to root-level access in version 7.4.

RHEL versions 5 and 6 are not as advanced, yet both still have plenty of major customers, Pacheco said. Of course, Red Hat is actively preparing version 8. An alpha version went to selected customers this week, according to an employee of the Raleigh, NC-based company who is not authorized to speak on-the-record. Officials declined to comment, however, it’s going to be a year or two before RHEL becomes mainstream.

RHEL 8 will draw heavily from the public Fedora distribution, which Red Hat sponsors. “We use Fedora as a technology incubation project, if you will,” Pacheco explained. An example is the current Fedora concept known as modularity. “The objective there is to be able to update a kernel separate from the user space, and vice-versa… we’re certainly keeping an eye on that,” he noted. That lets customers select their own combination of user space and kernel based on application requirements. Red Hat also expanded its reach in the containers world with the acquisition of CoreOS in January 2018.

With the rise of Docker and Kubernetes, the better question is ‘does it even matter anymore?'” Pirtle said. “If my entire platform orchestration is distribution-agnostic, I’m probably going to go with the lightest, cheapest, most secure OS available… and not be so concerned with vendor support contracts. I know I’m not alone in thinking that the operating system in the server farm has never been less relevant than today.”