Where I work we have Ubuntu for our servers - web, file, and application. When I asked why, I was told "because it was like that when we got here". No one knows why, we can only guess. I personally don't see why they would use Ubuntu for a server, I'd sooner use Debian than a Debian derivative. Given my preference, I would sooner use Slackware because I've used it for many years, and currently use it for a half-life, web, ftp, and minecraft server. The OS has never once gone down. But there is a huge difference between a server for games, and a mission critical server in a production environment.

So my question is, how suitable is Slackware to being used as a server in a mission critical environment? Are there tweaks that would make it more stable, perform better, or more secure? Or is it adequate as is? Or maybe I should ask, what should I do to make Slackware a good choice for a server in a mission critical production environment?

Honestly, if it is for a company, and it is mission-critical, I would use SLES or RHEL. Paying them for support is a good thing, especially when a critical server is down. You can then get immediate tech-support help. Granted, Linux is pretty well supported by the community, but pay-for support/debugging is better (in my opinion), in a corporate environment. Plus, things are often 'certified' to work with SLES or RHEL...like Oracle as a for-instance. Yes, you CAN run it on other distros...but Oracle support will say "Well, you're not running in a supported/certified configuration...sorry!".

Just my $0.02 worth. I have used openSUSE and Fedora before on servers, but ones that weren't mission-critical, like my NTP servers. I had four of them, so if one died, the NTP requests would still be easily handled by the others. Our network monitoring/ticket system, using Oracle back-end? RHEL, thanks...

Ubuntu doesn't always enjoy the best reputation around here, but consider: It's based on Debian (which was itself recently voted "server distro of the year" here at LQ). Ubuntu 12.04 has support through 2017, that's longer than a lot of distros (including Debian itself) can promise. Also Ubuntu is supported by Canonical corp, which offers paid support, a hardware certification program, etc. So Ubuntu is not necessarily a terrible choice for enterprise/server as you might assume or have heard rumors.

I don't know if a few local town halls, public libraries and a private school count as "mission critical environments", but servers and desktops are running fine here. To quote our mayor: "Since we moved the server to your Linux thing, we more or less forgot we have a server. We just use it."

I tweaked the Slackware 13.37 file server to make it more stable though: I moved it from the basement to the first floor, so it doesn't get flooded in case of heavy rainfall.

1 members found this post helpful.

+1 on that. I had a curious case this summer when Halliburton refused to support their software because it was installed on CentOS and not on Red Hat Enterprise Linux. I ended up nuking the install, purchasing a RHEL license and reinstalling everything.

I've been running a Slackware server in a university environment for (hmm) 17 years now! Web server, mail server and samba server. Never had any major problems. Very reliable. Maybe not mission critical in the corporate sense but my students sure rely on the web server.

1 members found this post helpful.

It is your comfort level. I would not use any of the *nbutu's in a mission critical environment. If I were to accept a position where such a setup was in place, it would be with the understanding I would be dropping the servers within 30 days of hire. I've had problems with crashes with Ubuntu, not something I would trust my data on.

I use Slackware as a server and would have no problem installing it in a mission critical environment. However, as noted above, Red Hat does have the advantage of direct support, at a financial cost. If I was running an Oracle Database, I would lean more towards what they recommend, i.e. Red Hat Enterprise.

-JJ

Certainly, Ubuntu is not the distro to use.

You can use whatever you like as long as it is stable. Indeed, many use RHEL, but it's hard to manage because it is usually outdated. I would use Slackware or Debian stable. However, if also depends on how you set it up. Make sure to avoid experimental software, and keep it as simple and manageable as possible.

Please remember that software in Slackware is as close to upstream as you can get. It's not patched heavily the way Ubuntu is. That goes for the kernel, Samba, and whatever else you might run on a server. So if Slackware isn't fit for mission-critical servers, then the Linux kernel isn't fit for mission-critical servers, and Samba isn't fit for mission-critical servers.

2 members found this post helpful.

I agree with you. For very mission-critical (for example telco, finance, ecc) is impossible think to have something not certificated and supportated.
Differently if we talk about public libraries, private school, or others where "normal" distro could go well.

When I watch the kids at our school hack away on the computers, desperately trying to a) grab some porn b) download movies c) install various games, I guess I could run a server for the NYSE anytime.

Don't get me wrong: I'm all about a 'free' distro (openSUSE, Ubuntu, Fedora, etc.), but sometimes it's better to be in a supported/certified position. Oracle cluster? I'll support it, from both Oracle and/or Red Hat/SLES. Internal network monitoring? If it's monitoring only, and only internally visible, I'd go free/unsupported, since if anything on it breaks, it'll be app related, not OS related. Just be paying $$$ for a "Well sir, we only support xxxx...we can try to help, but..."

I haven't used it before... is the professional consensus that Ubuntu's support and certification programs aren't on par with Red Hat's?

Doesn't "mission critical" have something to do with... what mission you're trying to accomplish? One thing I've noticed about Slackware -- over many years and versions -- is the stability and reliability; my servers (including the one I'm typing this on) run for months (in one case a year-and-a-half sitting in a closet mumbling to itself) without problems. I had to shut everything down 11 days ago because the power went off in a storm for four hours but prior to that everything had been up and running for months; the UPS gave up about two hours in (and two hour of warning beeps was driving me nuttier than I already am), so down we went.

These guys are running as data base servers (MySQL and PostgreSQL), not massive (certainly not Oracle) but fairly busy 24-7 along with geographic map server, file server, print sever and other activities and they just work. Oh, I do have to install patches when they become available but most of the time installation doesn't require a reboot (maybe stop and restart a service but not a reboot).

I've always found that irrespective of operating system the things you have to worry about are the electro mechanical gadgets (like disk drives) going blooie on you and it's a good idea to run RAID (and rsync to a back up server) but I sure value stability and reliability far more than bleeding-edge whiz-bangs and graphic nonsense (fine for desktops, maybe, waste of space on servers) and I don't like almost daily "upgrades" such as come piling in Ubuntu and some other distributions. OK, not daily, but certainly far too often for my tastes..

I get stability and reliability with Slackware. I don't deal with downtime. I like that. And, hey, it's easy enough to grab a box, install everything you need and let 'er run for a while and see what happens (or, better, what doesn't happen, eh?).

Hope this helps some.

1 members found this post helpful.