A CIO's View of Ubuntu

onehitwonder writes "Well-known CIO John Halamka has rigorously tested six different operating systems over the course of a year in an effort to find a viable alternative to Microsoft Windows on his laptop and his company's computers. Here is CIO.com's initial writeup on Halamka's experiences; we discussed their followup article on SUSE. Now CIO is running a writeup on Halamka's take on Ubuntu and how it stacks up against Novell SUSE 10, RHEL, Fedora, XP, and Mac OS X, in a life-and-death business environment." For the impatient, here's Halamka's conclusion: "A balanced approach of Windows for the niche business application user, Macs for the graphic artists/researchers, SUSE for enterprise kiosks/thin clients, and Ubuntu for power users seems like the sweet spot for 2008."

Having your cake and eating it too ...

[timholman]

For the impatient, here's Halamka's conclusion: "A balanced approach of Windows for the niche business application user, Macs for the graphic artists/researchers, SUSE for enterprise kiosks/thin clients, and Ubuntu for power users seems like the sweet spot for 2008."

Sweet. And with my Macbook and a copy of Parallels, I can have them all.

That's the beauty of virtualization on the Intel Macs. You cease worrying about which OS is the best compromise; you simply use the best OS for the task at hand.

Re:Ubuntu? Power users?

[cerelib]

Ubuntu still contains most of the command line maintenance utilities. So if you learn how to use them, you can do remote administration. On the other hand, as long as your network latency isn't horrible, you can use the GUI tools remotely. This can be done using either VNC or X. I use X clients remotely all of the time from my Windows laptop using Xming, an X Server for Windows. Just make sure you use port forwarding in your SSH session and you are good to go.

Re:A genius!

[good soldier svejk]

And I wouldn't think SUSE and Ubuntu are really all that different from a support perspective. Not sure why he thinks OSX is better for researchers, though. I tried looking at the article for more information, but I'm not going to wade through 17 pages of ads...

Having been part of this evaluation process, I can tell you that Ubuntu is much easier to support, but Novell offers far better enterprise support (including developer resources) for Suse, which is more important for the applications he proposes. I won't speak for John, but my guess is he thinks OS X is better for researchers because it it runs all the unixy apps the researchers require and even in its most wild and wooly user installed form is easily supportable by our existing resources. You can read the first article for more information. As John points out in the article, we have no control over what researchers buy with their grant money anyway. Except for a few "power users" who prefer GNU, there is pretty much concensus among researchers that OS X is the best platform for them. At any rate my experience here has been that there is no net cost to supporting OS X since our marginal cost for supporting Macs is lower than Windows boxes. OTOH, it probably isn't as good for kiosk workstation applications because of the lack of low end hardware options. In that application, where distributed support is a small fraction of cost, the best route is to keep capital cost to a minimum, which means GNU.

If you don't want all the annoying ads, click the "print" link and read it on one page. That is what I did.

Re:A genius!

[R2.0]

"I'm not sure I'd trust this guy as a Linux expert, however "well-known" he may be."

Sigh...

The whole point is that he is NOT a Linux expert, just like the other 99% of us out here in userland. Just like 95% of us are not Windows "experts". Allow me to clue in the 1%:

1) I don't care about KDE and Gnome either, nor do I care to know.
2) I don't want to be an "expert" at either system, but that doesn't mean I can't form opinions about how well something works for me or my organization.

It sounds like the Ubuntu folks seem to get what a large portion of the Linux community refuses to see - most end users don't care about esoterica. We just want it to work reasonably well. Not even perfect - just reasonably easy to use. Hell, I'm ready to make the switch to Ubuntu, but for my slavery to Quicken. But the other distributions? Meh. I ditched the command line with Dos and Win3.1 - my memory s crowded enough without having to emmorize command line switches for operations I don't do every day.

Re:Can anyone confirm?

[TheRealMindChild]

I don't know about evolution specifically... hell, my little blurb is coming from a windows world, but I figure programmers are programmers and they tend to make the same mistakes.

For example, if your firefox directory is read only, it takes MINUTES to fire up. Allow write access, it loads in a handful of seconds. Doing a little digging, it seems it is trying to open all of these config files for read/write... and when it fails, it tries a few more times. Then some of them get copied to $temp$ so that they CAN be opened for read/write, even though YOU LIKELY WON'T EVEN BE WRITING TO THEM. All it would take is a "if CantOpenConfigFileWithReadWrite(...) OpenConfigFileForReadOnly(...);"

And I use firefox as an example, but just about every application seems to have the same issues. This may be where Evolution is at.

Re:A genius!

[Otter]

The whole point is that he is NOT a Linux expert, just like the other 99% of us out here in userland.

No, but he's a CIO publicly holding forth on the suitability of one Linux over another for certain applications based on the failure to understand that you can change the desktop environment! Maybe I'm a Linux snob but that seems like a striking lack of understanding. It's not like he was complaining about the lack of some obscure functionality and I chimed in with "its fixed in CVS so stop spredding FUD you M$ a$troturfer"!

Re:A genius!

[Ichoran]

As a researcher, I think it depends on the field.

If you need to run specialized commercial software for data capture or analysis, you need Windows. Very few companies support anything else. Those that do (e.g. National Instruments) offer only a subset of their tools which aren't well integrated into the platforms.

If you just need a computer that is pretty and powerful and you don't have to worry about, you need OS X. Stuff just works; you can forget about the computing and focus on the research.

If you are in research that involves computation or statistics, you need Linux. The standard tools are more powerful and flexible than anything you can find under Windows, and the headache of getting these to work on a Mac more than offsets the slightly smoother interface in some areas.

And from what I've seen, researchers' preferences in these fields tend to follow the needs above. (People who are mostly interested in data collection/hardware interface generally prefer Windows, biology researchers generally like Macs, bioinformatics folks like Linux, etc..)

Re:This is not a job for a CIO

[Ichoran]

Supporting four platforms and getting them to play nicely together if you are starting with a large base of one platform where everything that works has been done without regard to using other platforms. Switching to a multi-platform solution would be a nightmare, especially when the original base is commercial ("Vendor lock-in"). As a business strategy, it wouldn't make sense to switch.

But supporting four platforms when you start off with that as your goal is not as much of a nightmare. We have the same thing where I am, and other than occasional recurring problems involving Windows not understanding that everything else is not also Windows, the support is actually better than if it were a one-platform area, since each platform is used in that area where it does best.

Re:A genius!

[Gromius]

Also as researcher (high energy particle physics, so very heavy on the computation and stats) I noticed that 50% - 80% of my colleagues own a mac laptop of some variety and that this is common though out the field. However they are used primary as an interface to our linux servers which are actually used to do the stats/computation. Apparently they are unixy enough having X windows and a terminal to be able to do this (unlike windows which always feels like a hack) while having that "It Just Works" ease of use. Although to be honest I often find that when interfacing with linux the mac is more like "It Just Works (well almost works, apart from a few fiddly things that you can probably learn to live without)" so I tend to avoid them myself.

Re:Mac's in research

[Frumious Wombat]

Not to mention that that extra dollar cost may push a Mac over the line into capital equipment where overhead isn't charged. My first laptop for work ran into that issue: I picked a nice model for ~$1800, and was told I wasn't spending enough. As it turned out, the $2600 ibm was cheaper, because the lower cost one came with 50% overhead attached.

You could have also ssh'd into a real cluster, or built a Mac cluster for a price similar to an Opteron system, and just quietly integrated it with your desktops (how my lab runs now). They just work, and they just work smoothly. It's also nice that tools like VMD come in native form, and run very smoothly. It's nice taking VMD [uiuc.edu], GAMESS [ameslab.gov], and Amber [scripps.edu] on the road with you, running in native mode the same way they run on the big clusters back home, just in case. (yes, I know about the windows ports or cygwin, but they always feel somewhat clunky)

Finally, sometimes the commercial software really does just work better, and fighting a journal over file formats is an exercise in futility.

Re:A genius!

[xenocide2]

The important question was how the two distributions performed without massive re-engineering.

When I read CIO evaluations, I expect a perspective that includes how organizations can create and deploy changes to the base platform. I know places that revert the Windows XP theme to "Classic", as it's more familiar to their people. If you're considering deploying Linux, you would do well to consider hiring a Linux expert to help you with such things, just as you hire Windows experts. Honestly, I understand your compatriot didn't write this article, but the level of detail offered is no substitute for expert advice. Perhaps there's a whitepaper report for sale in the works?

There are some important problems to recognize, although I hope you can pardon my amazement that people still want to listen to MIDIs at all. MIDI playback in Ubuntu is not as simple as it could be. While I don't know why this matters as an evaluation of 40,000 user base suitability, it might be the best example for the state of Ubuntu usability. At the moment, MIDI is recognized as a music filetype by GNOME, but gstreamer (and totem as a consequence) can't handle it. So first instinct when something doesn't work is to check the repos. There are 87 hits for "midi" in my apt-cache search. Once you exclude the libraries and random extra hits for midi maze clones and the like, you get about ten options. The first one is kmid. kmid looks like it would work out great in kubuntu, but I'm guessing it can't handle the lack of artsd running in the background or something, as I heard no sound. The last one on the list is timidity++. It works fine on the command line, but even if you install the extra interfaces, the interface isn't that great.

Gutsy (to be released in October) handles it slightly differently. If you double click to open a .mid, by default it opens up an install applications dialog, suggesting amarok or kmid. Timidity is tragically absent, and kmid still doesn't work after installation. Ideally, midi playback should be part of the gstreamer set of plugins, and MIDI would work out of the box with the default totem GUI. In practice, work has been done in gstreamer that basically ports timidity to the gstreamer framework [freedesktop.org] (as well as wildmidi, another midi library). This work was started in February 2007, so I can understand why it didn't make it into the current release. The better question, and one I don't immediately have an answer for, is why it's not yet hit development branch in Ubuntu. There exists a bounty to bring this functionality to life, so if anyone's looking to earn what appears to be around 200 dollars [launchpad.net], this whole problem could be wrapped up by October or sooner.

As an aside, I do appreciate the implication that Debian is the mother of all Linux. And we should recognize that organizations, hired bounties, or outside firms like SuSE, can make these re-engineering feats simple via open source.

Re:A genius!

[Anonymous Coward]

"Apparently they [Macs] are unixy enough having X windows and a terminal to be able to do this..."

And if you install MacFUSE from Google on your Mac, you get GUI file navigation/access via SSH on those Linux boxes.

Frequent crashes? Updates?

[RazvanHrestic]

Oh come one people, this seriously cannot be the real reason he wanted to switch OS. From the TFA and the introduction in CIO Magazine about the root cause of this change I seem to gather that John was annoyed with the updates that installed themselves and antivirus updates and so on. Any of these can be turned off or set to be manually installed later, and Windows instability? Yes, when you load it up with a hole bunch of applications, some legacy, some from vendors who don't know how to properly integrate their product with the OS, you might get some instability. Resolution to this sort of problems lies in application virtualization (see SoftGrid) or Terminal Services, or Citrix, which does incur bigger support costs, but maybe not as large as supporting >=4 OSes.

I cannot seriously see from the guy's description or even the CIO Mag's a real problem with the OS. I'd rather put this on account of his bias (also mentioned in TFA).

Please note that this is not in any way a bash of Ubuntu, SuSE, OS X or any other OS mentioned. I agree that they are more fit to do some jobs better than others. Hell, I even run Hoary Hedgehog on my old PC (converted to a sort of media-center). It's just the arguments are dubious.