Slashdot Mirror
Businesses Choosing "Community" Linux Distros
An anonymous reader sends along a PCWorld recap of a new study by the 451 Group, which claims that business use of 'community' Linux distributions is on the rise — distros like Ubuntu, CentOS, and Debian, as opposed to "corporate" packages like RHEL and Suse. The trend is most evident in Europe. The article points out examples in Sweden and Germany, and cites growing in-house expertise with Linux as one factor helping enterprises get comfortable choosing Linux distros without commercial support. Interestingly, the Swedish company mentioned, Blocket.se, has made a one-off support arrangement with their hardware vendor HP: "HP is really providing device driver and utility support it uses for customers running RHEL, but because the two distributions are binary-compatible, that support approach works just fine for CentOS. Blocket relies on its own engineers, systems administration, and software development to get its applications running on Linux. "
In Brazil, some times companies use Debian as their main SO, and hire their own support.
Grey's Law: Any sufficiently advanced incompetence is indistinguishable from malice.
How is Ubuntu not a corporate distribution? There is a
corporation developing and releasing that
product, even if it is loosely based on Debian.
This is how things are supposed to work with linux, isn't it? You support your local economy by using local people, instead of sending money away to whereever the HQ happens to be.
I thought this was one of the strengths with linux. Let's see if RH or SUSE has a business model that works according to this reality.
You are not entitled to your opinion. You are entitled to your informed opinion. -- Harlan Ellison
What I'm seeing across Europe is a growing use of not ABW (Anything but Windows) but WIW (What I Want). So developers are using Linux, and supporting it themselves, and execs are using Macs. A very common pattern is to see the "standard" corporate image run inside a virtual machine which gives access to the corporate email and other MS apps while the user spends lots of their time in the native machine doing their work. As a way to do "home working" this also works well as it means the corporate contamination of your home machine is limited to just the virtual image.
With more and more things being browser surfaced the need to have an MS box is reducing and people are choosing to use what they want and support it themselves. The corporate desktop therefore becomes virtual.
Personally I've a Linux laptop for Dev and a Mac OSX for the rest of my work, the Mac runs a windows VM for my corporate access.
This isn't a big religious thing its just that it works.
An Eye for an Eye will make the whole world blind - Gandhi
Ubuntu support contracts are available, same as Redhat, which is the reason we use ubuntu as our standard server, not debian. The other reason being our in house engineers are more likely to have ubuntu experience than redhat, as it's free and ubiquitous.
What about the Red Hat business model? (A little arm chair CEOing here - clearly I'm not CEO material, but this is Slashdot.) Hopefully, it can continue to support a steady stream of businesses migrating away from Microsoft for some time. But what about when that runs out?
The self supported businesses will still need to obtain their in-house expertise somehow. So training and certification would be one profit center. Contract work like IBM does would likely become the core business. Having an inside track as the distro maintainers is a valuable selling point, so continuing RHEL is vital - but must now be subsidized by training and contract work.
isn't that better for the economy overall than paying private company x for a complete solution. At least doing it this way keeps money and jobs nearby.
Jonathanjk.com
openSUSE is also a community distro where Novell is part of that community (as well as the sponsor).
Don't fight for your country, if your country does not fight for you.
Well, SuSE's distribution was always aggravating to those who wanted it for free (no free ISO downloads back in the day meant it was hard to install for free). RH was amenable to first-party free distribution until RH9, after which they decided this was their way.
Namely, both SuSE and RHEL have a 'commercial-only' distribution with those enterprise sensibilities and a free 'first-party' offering that is ostensibly an enthusiast endeavor which really translates to recruiting enthusiasts as testers. They bank on trademark/copyright of text and images to keep clones from looking *too* much like their first-party offerings. CentOS is from a technical standpoint, a clone (plus some other stuff, but the clone-only behavior is default), but distinguishable enough to preclude Vendor and ISV support (both don't want to go the linux support path alone generally).
Meanwhile, here comes Canonical. They truly keep the distribution and support model independent. They have rapid release cycles, but denote a more 'enterprise-friendly' LTS cycle underscoring things. Regardless, the distribution is free to download and distribute. So clients can prototype and train and even do production as they feel comfortable with doing so without support, and then when they do need support, the contract is available without reinstall or other drastic measures. Suddenly, the mark of whether another party will support it or not is not keyed on the distribution, instead requiring a Canonical support contract to be in place.
I think SuSE/RH's approach is botching the market. I know of a *lot* of CentOS installs going in to places that might feel more comfortable with the option of purchasing a support contract. Knowing the strict distinction between RH and CentOS, Ubuntu will be very appealing to those places. The absolute identical nature of free training/development/prototyping systems with low support requirements and production use is also appealing.
XML is like violence. If it doesn't solve the problem, use more.
We use CentOS on pretty much all our 150-odd Linux servers, except for those that require RHEL to be in a supported configuration (Oracle DB, Oracle Appserver, Oracle Financials).
Of course, while we mainly do this to save money, out of the million-plus we pay Oracle, the few thousand in RHEL licenses doesn't even count as a rounding error (hell, compared to Oracle licensing, even the cost of the hardware is irrelevant).
Does this present a problem in terms of one of the models of open source? One of the things often discussed on /. is the question of profiting from working in open source.
What's often been suggested is that there's money in support, and that if you create some software, and have experience then supporting it, that you gain a competitive advantage. That the likes of RedHat, MySQL etc will be customer's most likely first port of call.
If companies are simply going to go to someone else, that then suggests that investment in open source software could go down...
We use SuSE Linux Enterprise Server (SLES) from Novell for many of our servers, and are very happy with how easy it is to maintain (a lease cycle for the hardware eliminates the need for upgrades). I would be extremely hard-pressed to even consider using a community edition for production servers - that corporate-level support is extremely important.
However, when it comes to the desktop, the community editions offer more modern features - Novell's SuSE Linux Enterprise Desktop (SLED), is several years behind the current Open Source SuSE.
If the linux desktop ever comes of age for the average user, SLED may offer a very stable, easy to use environment (at least for supported hardware). However, since Linux Desktop is still primarily a developer's game, the OSS version offers the bleeding edge developers like, and know how to cope with.
"The large print giveth, and the small print taketh away" -- "Step Right Up", Tom Waits
Actually, there are plenty of reasons to choose a supported distro, even at a purely technical institution. For many organizations, it doesn't make sense to devote time or personnel to debugging system problems, and it often costs more to have an IT department handle everything than to have someone from Red Hat or Novell solve the problem. It is not just a question of hardware, it is also a question of software bugs, configuration problems, etc. Yes, any competent IT shop could take care of this, but that means devoting time that could otherwise be spent on business needs to solving little trivialities.
Or did you think the most successful financial companies in the world made an unplanned decision to pay Red Hat, Novell, Oracle, Sun, etc. large amounts of money for technical support, when they could have just done it in house?
Palm trees and 8
I could see this happening for smaller companies, but for the larger companies, I can't see them switching over. Large companies *hate* change. And I'd imagine that it's the larger companies who are using the corporate editions, while the smaller ones feel comfortable with the community editions.
Most of the businesses I've worked for I've pushed Debian as the distribution of choice. The biggest problem I see in mixed shops with Linux is often times there is no standardization on a single distribution. The one company I worked for had: Slackware, Gentoo, Redhat, SuSE, and some custom homebrew... I spent 3 months standardizing everything over to Debian. I built a standardized install manual, made sure we had a repository up to date with the latest drivers for special hardware, and setup all kinds of custom system status tracking with cacti and snmp. Management liked the new system setup so well that they eventually got rid of all the windows servers except two who ran custom software that our company's programmers wrote years ago and we lost the source code for.
Debian's free, the support is spectacular, it's package management is *excellent*, it's upgradable, it's easy to manage, and it doesn't install a lot of junk that is unneeded.
I *hate* rpm. It makes me crazy.
Shadus
I call astroturf on the above...
Actually, there are plenty of reasons to choose a supported distro, even at a purely technical institution. For many organizations, it doesn't make sense to devote time or personnel to debugging system problems,
Looking at the cost of labor, when you're working with low end stuff, it's usually cheaper to replace the hardware with something that is supported than waste labor time. When working with high end stuff, someone's job is/was on the line when they specified the equipment, so presumably they got it right due to careful research. It's a good question if there is a middle ground anymore or if that has been overlapped and eliminated.
and it often costs more to have an IT department handle everything than to have someone from Red Hat or Novell solve the problem
Usually the more people you involve the longer it takes. Realize that it is extremely unlikely that RHEL or Novell has hired author of the software that is having a problem, and probably not likely they have anyone with more experience than your own guys in your field of endeavor. It is also highly unlikely that you are having a problem with the distribution mechanism itself (bug in dpkg or apt-get or whatever). So, what it boils down to, is it more efficient for someone familiar with your local system to use google to find the answer, or to have your guys spend extra time explaining the problem to someone else, who knows nothing about your system, so they can use google to find the answer?
Or did you think the most successful financial companies in the world made an unplanned decision
Considering that virtually all financial companies are either bankrupt or going bankrupt due to fraud and stupidity, looking at them as a role model seems about a decade out of date.
Virtually all decisions made to buy support contracts are either:
1) Out of touch "pre google era" PHB decision
2) No internal skillset for something that is business critical, terrible is better than nothing at all.
3) Cascading interlocking licenses and requirements (you "need" oracle, which requires RHEL, so you "need" a contract) That is a bad economic structure which will eventually be worked around or eliminated.
"Science flies us to the moon. Religion flies us into buildings." - Victor Stenger
Once the top staff notice that they using the computer for work rather than spending all their time fighting Windows, you can probably zap that last box, too.
Most of the staff managed without any prompting from us. We were prepared for a lot of hand-holding that never materialized. Even with OpenOffice there hasn't been much. One question on how to do mail merge, I think.
The XP box in the flex area is supposed to be for guests and one of our vendors uses GoToMyPC for demonstrations and that doesn't work with Linux...that I know of anyway. And, yes, that's one of the vendors we're phasing out.
There is entertainment value in seeing the XP box sitting alone and unused in the flex area. Ultimately suffering the indignity of becoming the pedestal for the flex area scanner/copier and being periodically borged with a live CD. Poor sad little Windows box, nobody wants it. lol.
Puppy got the nod because it looks nice. I know that's not a great reason but if that smooths over the transition, fine. The laptops aren't that old. They have 256 meg of RAM and are pretty zippy running Puppy. The sales and execs probably use their Blackberries more than the laptops anyway. The only people with desktops are administrative, developers and support.
That's our life, the big wheel of shit. - The Fat Man, Blue Tango Salvage
Several years ago we here at NASA replaced Solaris X86 with Red Hat Linux as the operating system for our PCS systems (Thinkpad laptops used as the crew interface in the Space Station's command and control systems) We are currently in the process of rehosting again, this time to Scientific Linux, a CentOS-like rebuild of RHEL done by the good folks up at the Fermi National Accelerator Laboratory. Certainly cost was a factor, but not the deciding one. From our perspective it is golden not to have to track how many laptops each of the various development groups (many of which are international) have it loaded on.
It seems like several companies are still trying the tactic of software exclusivity, the same tactic the console companies are waging on one another. (In that arena, it's pretty unfortunate, too, as a lot of it just comes down to how much money you're willing to pay for exclusives, and Microsoft has the deepest pockets, or so their accountants claim.) This is something that cannot and should not occur in Linux as it hurts everyone. Part of software freedom is software accessibility, so when a new driver is created for example, it needs to be modular and easily pluggable into any Linux or Linux-like kernel, quickly and without hassle (the point of modules). Some companies are going to have to face the fact that they cannot get away with attracting everyone to their platform just because they have a certain software title, or just because they have large repositories.
Linux should be Linux, period. You should be able to use the entire Internet as your Linux repository. If package managers want to keep these so-called "third-party" packages separate from the ones they officially support for support contract reasons, so be it, but do not take away my freedom to install any piece of Linux software I want easily on any Linux distro. Cross-distro Linux packaging is more than possible and should become a reality soon.
So, without these "exclusive" distro-specific software packages, what remains to define a "distro"? Well, of course it's what it was from the start, a simple bundle of software for the convenience of being able to find all the basics, or simply the software you want, in one place. Linux distros should never be anything more than software bundles.
Help with Linux defragmentation. Support more standard APIs for desktop and general Linux interoperability to give everyone more choice and thus more freedom.
Promote true freedom - support standards and interoperability.
I work in small office.
We can't fit an IT staff member, let alone an entire IT department. There's a fellow doing "IT" but he works with our office and a few others that we work closely with.
Although shit never hits the fan (but I'm waiting for it any moment because of bad decision making by PHBs) and I'm able to resolve a lot of minor problems (they're a windows shop so sadly my experience is just helpful in trying to find a solution by exploring) I do believe a paid support contract would be worth it.
But then again, shit never hit the fan. So why should I pay for nothing?
I should have been more clear, I was referring to large IT shops that must guarantee 24x7 availability for thousands or even millions of users, such as the firms down on Wall Street. Trying to accomplish this in-house is often more expensive for these companies than paid support from Red Hat or Novell.
Palm trees and 8