Slashdot Mirror
New Debian-based Enterprise Linux?
arkanoid.dk writes "Sources close to Progeny, Mandriva and Turbolinux report that a new Enterprise Linux distribution is on its way. Apparently, the distribution will be based on Debian 3.1 Sarge and will form the foundation of the next server distributions from the three companies. The three companies hope that the new distribution will enable them to compete with the market leaders Red Hat and Novell Inc's server distributions. An interesting part is that the new system should support both DEB (Debian package) and RPM (Red Hat Package Management) to enable better cross-compatibility with other Linux flavours. The vendor said: 'It will have a nice, Web-based front end for service management, which Sarge lacks. It's basically oriented toward edge-of-the-network type applications, such as ISP software.'"
The reason my employer is spending big bucks on enterprise Linux (well, not quite so big now that we switched to SLES9) is simple: Oracle (the company, not the product) supports RHEL or SLES, and nothing else. Sure, we could run our databases on Gentoo if we wanted -- it works fine in practice -- but Oracle's terms prohibit that in a production environment, and we'd rather like to continue being able to use the expensive support we're paying for.
If these folks get Oracle certification, I'll be thrilled to have them in the marketplace. Otherwise, as far as my employer and I are concerned, they're "enterprise" in name alone.
(Oh -- and if you're considering Oracle, count the extra cost of a certified OS in as part of what you'll be paying for it. That, and the hair loss and headaches).
> And even if they were, is it really such a chore
> to upgrade every 12 months, when it's Free?
Yes it is. When they have critical systems that work, they don't want to break it under any circumstance. They don't care about what's new and hip (like geeks, consumers, and programmers like you and me). They just want to get work done and they don't want any surprises.
Most enterprises prefer a 3-5 year upgrade cycle.Some don't want to upgrade 20 old COBOL code, because it works and an upgrade would mean that you have to restart your QA from ground zero. All software has bugs and limitations, but with new software, you have a new set of bugs and a new of limitations (like new memory requirements or dropping old hardware devices). The key difference is that with the old OS, the limitations and bugs are known and workarounds are documented, while in the new OS they aren't. That's why the Linux 2.0 kernel is *still* being patched with security fixes, even though it was first released in 1997 and is really stale. That's why Windows *2000* is still quite popular in enterprise even though everyone knows that move to XP/2003/Longhorn is inevitable, and why some enterprise software still requires Windows NT.
If it works and security patches (without new features/bugs) are kept up to date, why should you have to "fix" it? The only thing Debian needs to be enterprise grade is a predictable 3-5 year release cycle (or the willingness to support all versions of Debian for this long.) so that enterprises can plan their upgrades and burn in verification in an orderly fashion. That appears to be what the new Debian president is promising.
The new "United Linux" group just adds a bit more credibility to that promise.