Canonical Chases Deal to Ship Ubuntu Server OS

Kurtz'sKompund writes "Canonical, the company that supports Ubuntu Linux, is trying to work out a deal with hardware vendors such as Dell to make Ubuntu available pre-installed on servers. 'Canonical, despite obviously supporting such a deal, had little to do with Dell's decision. Dell said it was merited by customer demand. Likewise, the decision of whether Ubuntu Server will ship pre-installed will be determined the same way.'"

Comments

Pre-installed OS

[totallygeek]

I wonder how many of you want pre-installed operating systems - any operating system. It seems to me that most installations have a central installation service (kickstart, jumpstart, etc) and would like boxes to be configured with certain parameters and niceties, such as partition and block sizing, configuration files, kernel versions, cfEngine keys, local accounts, etc... I really am curious, as I have never used the factory-installed operating system. Even if it came with one, the first item on the 'to do' list was to wipe the drive and start over.

Re:Pre-installed OS

[Hercynium]

Well, while I tend to go thru the same exact process on every server I deal with, I can see *one* big advantage - Reputation.

The typical PHB has by now recognized RHEL as *the* Linux for servers. Thru good marketing, development, support, and business, RedHat has become the de-facto standard for Linux in the enterprise server market.

Case in point - Not ONE of the enterprise apps I work with is supported on anything BUT RHEL, (or in one case SuSE) HOWEVER, I've tested many of them in the lab with Debian and Ubuntu and found that all work very well... but there's a snowball's chance in hell that management would let me use Debian or Ubuntu. RedHat's reputation as Linux for serious business is entrenched in their minds, and entrenched in the market.

I have a lot of respect and appreciation for RedHat's offerings. I prefer Debian, and in the corporate world, Ubuntu is the only Debian derivitave that has a chance of becoming a contender.

Being a default offering on Dell servers is a golden opportunity to start building the reputation they need. PHBs will see the Ubuntu option on Dell's web-site and after about a thousand times they may begin to wonder if it's something worth investigating. :)

If Canonical produces a systems-management/data-center platform that can compete with offerings available for RedHat, I believe that sysadmins, enterprise software vendors, and even managers will start to take notice. If Ubuntu can garner reputation as an alternative to RHEL, we may start seeing not just hardware support but also software support.

Granted, this is all just a wild dream for me, but let me tell you - if someday Oracle announces support for Ubuntu, it could be a dream come true!

the ballmer effect

[User 956]

Dell said it was merited by customer demand.

In other words, "No, Microsoft, we haven't been talking to other OS vendors. It was the customers' fault. honest. Put down that chair."

Re:Worst-case-scenario for Linux as a whole

[radeex]

1. Actually, Canonical offers professional support services for servers and desktops. http://canonical.com/support

2. My impression is that the "gaps" referred to in the article are mostly about certification from third parties like Oracle.

You guys will have to seriously take it up a notch

[outZider]

As a user of Ubuntu Server on a 60+ machine deployment for $work, Canonical is seriously going to have to take the server distro up a notch before this could possibly work. The server distro, in stark contrast to Desktop, is a horribly hacked together mess that gives off the impression that it isn't really studied much at all. While Desktop shows the best and brightest of Linux integration, the server distro is just as barebones as the alternative distro, and manages to screw things up in terms of out of box experience.

Am I looking for a UI? No. I want a few basic things.

1: A proper, usable deployment system. debian-installer is good for the basics, but it's a pain in the behind to set up, and doesn't support scripting a RAID/mdadm install, or LVM. This "sucks". Take a look at Redhat or CentOS for a little inspiration.
2: A boot screen that doesn't look like vomited output. Why does the login prompt appear before services have finished loading? I support being able to use the machine before services have stopped. I do not need "Starting PostgreSQL" appearing as I'm entering my login credentials locally.
3: A server kernel that always installs. Why does the installer give me the generic kernel when I'm installing the server distro? Why do I have to manually install the server kernel on boot up, and then remove the generic kernel?
4: Easily add services. You get 'LAMP server' or 'DNS server' or nothing. I had to create a custom installer just to have openssh-server install by default on first load, without apache or MySQL, or other crap floating around in there as well.

It sounds whiny, I know, but we really like the debian-style package management system with the modern services Ubuntu provides. It's great for that purpose. As a real server distro, though, long way to go yet.

I hope this lights that fire under Canonical to pay some attention to Server.

Re:Why Ubuntu?

[Shados]

I don't disagree with you, but one has to realise that servers are not all the big mission critical machines in datacenters we tend to picture them as. There are probably as many "non-mission-critical" servers as there are the others, and for those, a "Server for Dummy" install probably is cost effective in the long run.

Case in point: the company I work for offers a relatively advanced web solution. The software doesn't actually deal with mission critical data, it is used for projections and on the fly analytic operations, on a user per user basis. So each user has a copy of the data and basically mess with it the way they bloody want until they get an acceptable result, print a report, then go to their primary system (which isn't by us, and is totally independant in every ways, shape and form) and perform mission critical operations THERE.

For our servers, we can toss the app on anything, passwords can be in plain text (well, could if users didn't reuse passwords all over, which isn't the case so I guess they can't!), the machine can be tossed and kicked around, it doesn't really matter if the system's down for a day, or a week, as long as it comes back and it "works".

This is actually an incredibly common scenario, and more and more as a lot of software is moved to simple web apps (because of the Web 2.0 overhype) and other such things, especially since hardware is so cheap (I've seen servers running cache engines made with less than 300 lines of code, including comments, in a farm... hardly mission critical either), so there's IS a pretty high demand for "dumb-friendly" servers that don't even require the sysadmin intervention when they screw up.

In such cases, something like Ubuntu Server probably fits the bill amazingly nicely. If the machine screws up BAD, you call the sysadmin...but the rest of the time, let said professional handle the important stuff, and have the junior manage the non-critical, novice friendly environments. Saves time and money for everyone.