I should mention that I'm a big Ubuntu fan and have used it for five years. Even
so, it pains me to see the obvious ways it could improve. As I'll
explain, I believe Canonical's business model holds Ubuntu back from
fulfilling its potential.
Why It Matters
One obvious response to anyone who criticizes Ubuntu is to say to them: why
don't you just run another operating system? There are so many
competing Linux and BSD distros out there.
True. But there is a larger issue here. Ubuntu's great popularity means that it represents Linux to many people. It's the distro vendors pre-install. It's the distro the mainstream media always review. It's the one distro everybody's tried. It's been ranked #1 in DistroWatch's yearly popularity ratings for the past six years (1).
Fair or not, Ubuntu reflects on the Linux community as a whole. How well Ubuntu meets criticisms matters even to Linux users who don't use it.
So what are common Ubuntu criticisms? Here are those I often hear...
It's BloatedTo say that Ubuntu is bloated only makes sense if comparing it to some alternative. So let's do that.
Is Ubuntu bloated compared to Windows?
This chart compares Ubuntu's system requirements to the last three Windows releases:
||Ubuntu 10 and 11:
||128 / 512 m
||1 / 2 g
||1 / 4 g
||512 m / 1 g
||$ 199 - 299
||$ 239 - 399
||$ 199 - 319
Sources: websites for Microsoft and Ubuntu, plus web articles and personal experience. Chart is simplified and details have been omitted for clarity. Microsoft offers many Windows editions, this chart addresses the most common. Microsoft prices are for full versions. In the Memory column, the first number for each system is generally considered the minimal realistic memory, while the second is the memory recommended for best performance.
By any measure Ubuntu is not bloated compared to Windows. I'm writing this article with Ubuntu 10.10 running on a seven-year-old Pentium IV with a single core 2.4 ghz processor and 768 M of DDR-1 memory. This computer wouldn't even boot Vista or Windows 7. It runs Windows XP great, but that's not current software. XP is two Windows releases back.
Is Ubuntu bloated compared to prior releases?
Ubuntu's system requirements indicate the product's resource requirements have crept upwards over the years. Here are its memory requirements:
|Ubuntu Desktop Version:
||10.04 / 10.10||11.04|
||512 / 1 G||512 / 1 G
These RAM requirements and the recommended minimum 1 ghz processor mean that nearly any computer sold in the past seven to ten years can run Ubuntu. I've run 10.x on P-IV's and even P-III's. By this measure, one could hardly label Ubuntu "bloated."
Is Ubuntu bloated compared to other Linux distributions?
Linux distros divide into full-size, mid-size, and lightweight. Ubuntu is full-size.
Most full-size distros come in multiple versions. Their standard product usually requires at a P-IV or better with at least 512 M to 1 G memory. You may be able to get by with lesser hardware but it's not recommended.
Mid-size distributions like the standard editions of Zenwalk and VectorLinux go a bit lower than the full-size distros. They'll run fine on a P-III with 256 M. Lightweight distros like Puppy or VectorLinux Light Edition will run down to 128 M or less if properly configured.
To compete with this, full-size distros usually offer pared-down versions for those with lesser hardware. For example, Ubuntu offers Lubuntu; PCLinuxOS has PCLinuxOS LXDE and other variants; Mint can run with lightweight GUIs like LXDE, XFCE, Fluxbox; and so on.
Compared to other full-size Linux distros Ubuntu is not bloated. For something lighter, try Lubuntu. Lubuntu requires half Ubuntu's memory and only 1/3 to 1/2 of its disk footprint. It's also lighter on the processor. Read my detailed review of Lubuntu here.
It Lacks Enterprise Integration
This complaint is that Ubuntu lacks the enterprise-wide integration and manageability critical to large
System administrators require a single control point for automated administration and monitoring of remote Ubuntu desktops. Landscape, Canonical's product for enterprise-wide management, fulfills this need. But it is too narrow to address the larger integration issue. What about a single sign-on for login, email, and web access? What about directory services? How about Kerberos network authentication and LDAP (Lightweight Directory Access Protocol) support? How about coordinated information management across client and server products?
Microsoft is the competitor in this space. Its full range of client and server products seamlessly integrate. The server products include Active Directory, Exchange Server, and SharePoint Server. Client products like Windows desktop, the Outlook email client, and the Office suite seamlessly integrate with the server software.
There are two ways Canonical can challenge Microsoft's client-server headlock on the enterprise. It can either:
- Directly compete with a full range of directory, mail, and information management services
- Better integrate Ubuntu desktop into the Microsoft ecosystem already in place at most companies
The second option is in progress at Edubuntu but not complete. It leverages standards like Kerberos and LDAP to facilitate integration.
One system administrator summarizes the situation this way, "... Microsoft continues to win on the desktop. Not because an individual PC running Windows is easier for most people to use, but because its easier to set up Active Directory to work with Outlook and Exchange than it is to roll your own directory service with the tools available out of the box on Ubuntu."
Here's a management consultant whose clients manage between 50 and 150,000 desktops: "Until there is a true competitor to Active Directory, Exchange, Outlook, and the MANAGEMENT of the machines, Ubuntu will not succeed in the Enterprise."
Too bad Canonical let Attachmate Corp. buy Novell when the company was up for grabs late last year. Novell products like eDirectory and GroupWise could synergize with Ubuntu. Canonical's Linux dominance plus Novell's directory services and deep experience integrating into the Microsoft ecosystem might have been very competitive.
Perhaps cloud computing will ameliorate the integration issue. Organizations may shift their integration focus from internal servers to cloud services. This is the premise underlying Google's Chromebook.
In any case, Canonical needs to recognize this key source of corporate resistance to Ubuntu and make explicit their plan to overcome it. Then they need to promote the plan in the IT community. Thus far they have failed on both counts.
It Doesn't Install CompleteHere's a complaint with which we're all familiar. Ubuntu bundles a ton of great software but leaves out some essentials. Codecs, Adobe Flash Player, multimedia players, and proprietary hardware drivers are examples. You can easily install the missing programs, but you have to:
- Know what is missing
- Know how to install it
- Make the effort to install it
Canonical addresses this criticism in several ways. They segregate non-free software into its own Multiverse Repository, so that it can easily be identified and installed. Medibuntu (Multimedia, Entertainment & Distractions In Ubuntu) is "a repository of packages that cannot be included into the Ubuntu distribution for legal reasons (copyright, license, patent, etc)." Users can check for proprietary hardware drivers through the Startup Applications panel or the Administration -> Hardware Drivers option.
Good documentation and How To's help Ubuntu users. But navigating these can be difficult for the inexperienced. Not all docs are dated or identify the release(s) to which they refer. In the worst case, the user googles and retrieves conflicting instructions for a simple task they want to perform.
Some distros build on top of Ubuntu to give a more complete user experience. Linux Mint, for example, states its first goal as: "It works out of the box, with full multimedia support and is extremely easy to use." PCLinuxOS is another competitor that emphasizes it is "a full multimedia operating system."
I feel the "completeness criticism" is but a nit for experienced users. They can easily install the few apps or plugins Ubuntu doesn't initially provide. For newbies, though, this is a hurdle. End users don't know and don't care about the debate in the Linux community over "free versus non-free." They just want software that does everything they want with as little effort as possible.
Here's how Canonical could address this problem. Add an install panel allowing the user to select what goes into his installation. Give him a checklist of installable products -- with each denoted as free or proprietary. Users could choose software conforming to the IP laws of their country. With the customer checking acceptance of licensing conditions, Canonical would be absolved of legal responsibility. Users would get the most complete system permitted in their jurisdiction by a simple install panel checklist.
It Doesn't Install SecuredComparative studies and vendors alike confirm that Linux has a superior track record as a secure operating system. Ubuntu upholds this great tradition. You'd be hard-pressed to find evidence of malware infections in the Ubuntu community.
But does Ubuntu install as secure as it could, right out of the box? Surprisingly, no.
Take the default firewall as an example. In version 10.x, the Uncomplicated Firewall, or UFW, installs as Disabled. You'd think such a fundamental security tool as a firewall would default to Enabled. Or failing that, that the installation panels would give you a checkbox for enabling it.
UFW's front-end management interface, Gufw, doesn't install by default. You get the firewall without the GUI to manage it! The user must know about Gufw and install it separately.
How about configuring the firewall? Windows products like ZoneAlarm help you "train" them. They intercept each program the first time it communicates through the internet, and ask you to Allow or Deny the communication. Then they automatically generate the proper firewall rule for your decision. They also provide a checklist of installed programs. You simply check Yes or No for each program, indicating whether it has Incoming and/or Outgoing Internet communication privileges.
In contrast, UFW expects the user to write its rules with its barren, minimalist GUI. This is neither state-of-the-art nor competitive. It's certainly not user-friendly. As a friend complained to me: "I don't want to manage ports, I want to manage programs!"
To anyone who claims that Ubuntu "doesn't install secured," I'd say the product's outstanding track record argues otherwise. This is a highly secure system. Yet ease of configuration is missing. This isn't the only area where Ubuntu's ease of use falls short...
Its File Manager Isn't User FriendlyEver taught a class of new Ubuntu users when they run into
Nautilus? They always ask how to create a sub-folder instead of a top
level folder in a filesystem. They ask how to copy folders to their USB drive or backup disk.
Nautilus doesn't always show that a copy worked as expected, and if you're overwriting an existing file, it doesn't display timestamps so that you know which copy is the more recent. It doesn't always display error messages. For example, try to delete a directory for which you don't have valid permission. Or copy into that directory. You won't get an error message! Users need feedback. The old Unix dictum "no news is good news" is completely inappropriate for products that target end users.
There's an easy fix. The huge Ubuntu software repositories contain more than a dozen competing file managers. Ubuntu's superior install tools -- the Ubuntu Software Center and the Synaptic Package Manager -- make it easy to download them. If you don't like Nautilus, just click the mouse a couple times and install another product.
The mystery is why Ubuntu bundles Nautilus as its default. File managers are one of the most frequently used tools in any operating system. Consumers expect to use the default file manager without having to replace it. Fixing or replacing Nautilus should be a no-brainer.
It Won't Run Windows SoftwareThose who make this accusation either aren't familiar with Wine, or they haven't used it lately. The Wine database lists over 16,000 Windows programs that it runs on Linux. I'm constantly surprised that even big, complex applications run under Ubuntu with Wine. Examples include web site generators like Adobe Dreamweaver and NetObjects Fusion, and office products like Microsoft Office and Adobe InDesign.
Wine works like you'd expect. After installing it, you run Windows programs in the exact same manner you would under Windows.
Another compatibility option is DOSBox, an emulator designed for old DOS software. I have a number of simple Windows 3.1 games, such as Ringo, Ludo, and Boule (free download here). The games run fine under either Wine or DOSBox. They don't run natively under either Vista or Windows 7 -- even with its new Program Compatibility panel. Compare Ubuntu with Wine and DOSBox to native Vista and Windows 7, and you'll often find that Linux is more compatible with old Windows programs than Windows!
I've found an analogous relationship between Microsoft Office and OpenOffice. Microsoft releases new versions of Office every three years or so: Office 95, Office 97, Office 2000, Office 2003, Office 2007, Office 2010. (This excludes MacIntosh versions). As far as I can determine, the company only regression-tests back one version. The result in my experience is that OpenOffice is often more compatible with older versions of Microsoft Office than is Office itself.
When critics complain that Ubuntu is not compatible with Microsoft software, I sympathize. In spite of all that I've pointed out, gaps persist. But when one considers Microsoft's own software -- rooted in a business model of continuous releases based on planned obsolescence -- it becomes apparent that compatibility is not an issue only for Ubuntu. Depending on your compatibility needs, you may get a better deal from Ubuntu than from Microsoft.
It's BuggySeveral academic studies and papers conclude that Linux and open source software have fewer bugs than commercial products. Ubuntu has bug-tracking identification and resolution procedures equal to those of any large, well-run software project.
From years of participating in the Ubuntu forums, I've encountered consistent anecdotal evidence. I read very few posts where a user abandons the product due to a bug. This is a huge vote of confidence in Ubuntu. (You can't say this about every Linux distro.)
However, it's not unusual to see posts from first-timers who abandon Ubuntu due to install issues. Examples are things like Ubuntu not recognizing a sound card, or being unable to get wireless networking going, or a display problem of some sort. While these may not be bugs, they are cases where Ubuntu doesn't work for the prospective user. If I were to recommend one area for the Ubuntu team to target for a better user experience, device recognition and configuration would be it.
A related issue is that Ubuntu actually removes hardware detection capabilities as new versions come out. So a machine that worked fine with an older release of the product suddenly fails when you move to a newer release!
I've maintained Ubuntu instances for five years, since release 6.06, and have repeatedly run into this problem. In several cases video worked fine on one release and then fails under a newer one. Right now I'm trying to fix wireless networking on a laptop that worked fine in 8.04 and fails under 10.04. It doesn't work whether I do an upgrade or a fresh 10.04 install. (Wireless works fine for this laptop with Puppy Linux and Windows XP.)
Admittedly, device recognition and configuration is a sisyphean task. When you try any Linux distribution for the first time, you just hold your breath and hope that the product recognizes all your devices. This remains Linux's biggest challenge.
From the user perspective, though, to have a product that works fine under one release break under a newer release... that really doesn't look good. If there is a single issue that tarnishes Ubuntu's reputation, comprehensive, consistent device recognition and configuration is it.
It Changes Quickly But Doesn't Protect Its Users
Ubuntu improves rapidly. In the last two years, the product has moved from the GRUB boot loader to
GRUB 2, to continually changing networking management tools, to eliminating
the xorg.conf configuration file and moving to RandR for video, to switching the user interface
from GNOME to Unity, to replacing OpenOffice with LibreOffice. I've read about replacing GDM with LightDM, moving to more regular updates, replacing X.org with Wayland, and more.
Ubuntu's aggressive improvements are among its greatest strengths. But this benefit causes work for the existing user base.
The Ubuntu team could easily shield their customers from the impacts of these changes. Often they don't.
Here's an example. With GRUB 2 you no longer configure the boot menu of OS options by editing the menu.lst file. Instead, you edit bash scripts. That's fine for me, but an unreasonable expectation for end users. How about a simple GUI front end for editing the boot-time menu?
Another example: new releases take away the xorg.conf video display file that generations of Linux support personnel are accustomed to editing. You can generate this file and then edit it if you look up the commands to create it. But why should you have to? Why doesn't the System --> Administration menu have a button to generate a xorg.conf file for you? And automatically plop you into editing it?
A final example. Right now I'm researching how to install the Java browser plugin under Ubuntu 10.04. Websites are providing conflicting answers. This was trivial in earlier releases. But no longer. Apparently we switched from Sun's Java packages to OpenJDK. Beyond inadequate details in the Release Notes, no one bothered to insulate the users from this change. Why is it put on the customer to manage this change?
The Ubuntu team does a superior job in adding new features. They need to protect their users from the disruption these changes cause. This should be a top priority because it deeply impacts the product's ease of use.
To the average consumer little GUI "transitional aids" like those I've mentioned would help tremendously. They would be trivial to program. Why doesn't Canonical include them? Is it simply a lack of focus on ease of use? Here's my theory ...
Fix the Business Model
Of the above criticisms, those I feel have the greatest merit focus on whether Ubuntu is as easy to use as it could be. You see this in:
- Device recognition
- Default file manager
- Security configuration
Consider Microsoft's business model. The company makes 27% of its total sales revenue from Windows and 27% from Office (2). That's over half Microsoft's revenue. Without it, the company as we know it would cease to exist. Microsoft can't afford to stick with a product and polish it until it shines. Its business model forces it to constantly update, replace, and repackage existing code into new product.
No Windows version achieves its full potential because Microsoft must abandon it to introduce revenue-generating new product. New features are critical because they are used to justify the new version to the consumer public. The GUI is often the focus of "improvement" because it is the most visible to customers.
The history of Windows releases verifies this continual forced march to new product:
Courtesy: Wikipedia article
Canonical implicitly accepts Microsoft's disruptive business model as the terrain for their competition. Ubuntu directly challenges Windows in the new features competition. And it succeeds. But other design goals get pushed to lower priority.
Here's an example. Canonical and Microsoft sell to both consumers and corporate customers. They drive product change from the consumer side. This conflicts with the expectations of their corporate customers. Corporate customers value stability, compatibility, minimal bugs, and ease of upgrades over the headlong rush to new features.
Canonical tries to bridge this gap through differentiated policy, support, and pricing. For example, they distinguish between Desktop and Server products, and between regular and Long Term Support (LTS) releases. They offer corporate customers comprehensive support options and contracts.
Readers with long memories might recall that Red Hat also got caught in the conflict between consumer and corporate expectations. The company flip-flopped several times over their support for desktops versus servers. Ultimately Red Hat solved the conflict by spinning off desktop Linux to the Fedora project in 2003, while it went forward with Red Hat Enterprise Linux for servers.
I believe Canonical would be better served by protecting those who find that rapid change causes them work -- its user base. Polish existing code to improve ease of use. Concentrate on easy upgrades, great device recognition and intelligent automated configuration. Minimize bugs. Abandon the pell-mell rush to new features. Improve the product at a measured pace. Nurture and organically grow the base. New users will come naturally if the product provides solid long-term value. You needn't hype an "all new" interface to attract them. That's Microsoft's game.
The best way to compete with Windows isn't to mimic Microsoft's business model. You win by presenting an alternative vision grounded in a unique competitive model.
And the Consensus Is?Ubuntu's popularity means that it represents Linux
to many people. How well the product meets criticisms is important even to
Linux users who don't use it.
I've presented my views to stimulate your thinking. But here's a better idea. Why don't we see if we can come up with a community consensus? Add your comments to this article to address:
- What is Ubuntu's greatest strength?
- Are any of the criticisms listed here valid?
- If you could ask the Ubuntu team to fix one thing or improve one area, what would it be?
- - - - - - - - - - - - - - - - - - - - - -
Howard Fosdick (President, FCI) is an independent consultant who supports databases and operating systems. Read his other articles and download his free guide How to Tune Up Windows from here. You can reach him at contactfci at the domain name of sbcglobal (period) net.
Footnotes(1) You can view historical distro popularity rankings at Distrowatch by changing the time period in the drop-down list box under the label Data span. You must press the Refresh button to see updated statistics.
(2) The 12% Letter, March 2011. The March issue of this investment newsletter analyzes Microsoft's business from an investor's standpoint.