Hardware 2.0

Adrian Kingsley-Hughes

Does it matter that "Linux is bloated"?

By Adrian Kingsley-Hughes | September 22, 2009, 7:03am PDT

Yesterday at LinuxCon, Linus Torvalds founder of the Linux kernel, made a few comments about Linux usually reserved for Windows:

“Linux is bloated!”

It gets worse …

“We are definitely not the streamlined, hyper-efficient kernel I envisioned when I started writing Linux.”

And …

“The kernel is huge and bloated, and our icache footprint is scary. I mean, there is no question about that. And whenever we add a new feature, it only gets worse.”

OK, harsh words, but does it matter? Well, from the point of view of the user I don’t think it makes any real difference. In fact, the bloated nature of the kernel might actually help Linux compatibility. Given the wide variety of platforms that Linux can run on, it’s hardly surprising that the kernel isn’t “streamlined” or ”hyper-efficient”.

According to Torvalds, the it’s not causing much problems with regards to stability:

“I think we’ve been pretty stable. We are finding the bugs as fast as we’re adding them, even though we’re adding more code.”

From a developer perspective a messy kernel will make things more difficult to update the kernel. But then again, in a situation where a project is constantly on the grow, adding new features is sexier and take priority over tidying up existing code. In fact, getting developers to go back to basics can be an uphill struggle.

But it is interesting to see how open source is vulnerable to the same problems associated with bloat in the same way that commercial projects are. What I think is interesting is how as Linux has become bloated, both Microsoft and Apple have been working hard at optimizing code and tidying up their code.

Bottom line though, unless code bloat is at the point where end users are affected by it, then it’s not really a big deal. Bloat from the perspective on the user, and bloat from a developer’s point of view are two very different things.

Kick off your day with ZDNet's daily e-mail newsletter. It's the freshest tech news and opinion, served hot. Get it.

More from “Hardware 2.0”

Has iPhone OS 3.1 battered your battery?

Can Microsoft duplicate the "Apple Store" experience?

Topics

Kernel, Linux, UNIX, Operating Systems, Open Source, Software, Adrian Kingsley-Hughes

Adrian Kingsley-Hughes is an internationally published technology author who has devoted over a decade to helping users get the most from technology.

Full Bio
Disclosure
Contact

Vendor HotSpot

The 360° Conversation around Cloud Computing

TECH VISUALIZER brings the social conversation to life... powered by Brocade

Learn More »

Related Discussions on TechRepublic

Did you know you can take part in these discussions with your ZDNet membership?

Linux command line tips: history and HISTIGNORE in Bash

Ask a Question Start a Discussion

Comments

Add Your Opinion

Join the conversation!

Follow via:: RSS; Email Alert

Just In

Bloat

bofaboy 22nd Oct 2009

The fact that the people involved with Linux are acknowledging the bloat is encouraging. the big non-Linux corps daren't address this topic . . for the best take on Bloat check out http://www.computerworld.com/s/article/9136280/Opinion_The_end_of_bloatware_The_return_of_programming_s_golden_age_

View in thread

Please Select
Please Select

0 Votes

+ -

well compare to what

Quebec-french 22nd Sep 2009

compare too its former self ... ultra thin but
lack finesse

compare to win 2k
compare to win xp
compare to vista /win 7 will all there crapware

compare to Mac and its hardware lock in

Its easy to state Linux is bloated but on
what ....

the vision the Linus add back then ....

Linux is bloated maybe in the end it change
what performance , usability , bloated what
.....

Reply
Flag

0 Votes

+ -

Maybe I am misunderstanding you

GuidingLight Updated - 22nd Sep 2009

So the fact that Linux is bloated is acceptable because Windows and OS X are bloated?

Reply
Flag

0 Votes

+ -

I think he's arguing that it isn't bloated

Michael Alan Goff 23rd Sep 2009

Because of the size by comparison.

0 Votes

+ -

Bloated Linux

Robert Wheeler 24th Sep 2009

Well, you know the x86, and maybe ppc is like this, but not nearly as funky.. Is full of gotchas.. and honestly not one os can cover everyone of these gotchas, such as drivers. While it would be nice to see at least one Linux distro, highly intergrated, gui and all. With a low bloat count. Yeah this would be good. But you have all these gui's slapped on a linux kernal, not one big project, no many small projects.. all trying to work togther. Yeah so.. it would be nice to see that. Whos going to actually make a lean and mean linux? Every single distro I have ever used still had some kind of bloat, and most of them where lacking most of the "features" I would find as needs. Yeah this is far and away the truth about most modern os's there bloated. Windows 7 uses a hibernation file by default, and theres no "easy" way of just turning off that feature... Mac OS X has crud too.. and You stop and think, Doesnt it really have to be? Not if we where all using the same hardware, and the whole os was developed by the same people... Even then, It's unlikely to happen!

0 Votes

+ -

compared to an OS sitting on a devastating exploit for 8 years

LBiege 22nd Sep 2009

... and doing nothing about it while blatantly claiming self to be secure b/c of many (blind) eyes are reviewing its code?

Reply
Flag

0 Votes

+ -

Compared to

PaddyO 24th Sep 2009

the null pointer exploit researchers recently found (while no one was doing nothing about it, by the way, since it was only recently discovered) in Red Hat's kernel and for which a patch has already been released?

The exploit that nobody has exploited since 2.4 because, well, it took those blind-eyed researchers to find and and back track it to 2.4 while nobody was actually, well, exploiting it?

That exploit that nobody has been doing anything about except for patching it as soon as it was discovered? The exploit that is fixed if people bother to update?

Do try to keep up, please.

0 Votes

+ -

BTW...

PaddyO 24th Sep 2009

Forgot to mention.

It affects local privilege escalation.

Physical access has, and always will be, the mother of all security risks.

That goes for all operating systems.

0 Votes

+ -

Well, duh. That's what you get with a monolithic kernel

frgough 22nd Sep 2009

model. It was a stupid decision when he initially made it, and
now it's come back to bite him in the butt.

Reply
Flag

0 Votes

+ -

Excuse Me ?

linux4u 22nd Sep 2009

Most Ditro's ship the kernel in a modular fashion. The Modules that are need will get loaded when the system boots up.

On the other hand, as I learned in MY CS courses back in th 70's, that the best OS is one designed specifically for that piece of hardware, SO my kernel's tend to be a MONO kernel custome configured for what I am running it on.

While try, there ARE probably WAY more modules / drivers /low level routines than what is really needed and it all adds bloat, but makes the kernel pretty much run on anything.

Reply
Flag

0 Votes

+ -

It has to be for performance reasons

T1Oracle 22nd Sep 2009

All kernels are monolithic because microkernels are too slow.

Reply
Flag

0 Votes

+ -

If microkernels were easy...

TripleII-21189418044173169409978279405827 22nd Sep 2009

...we would already be using them. They are incredibly complex and even the most basic examples have an overhead in their processing that kills performance. Where a monolithic kernel can pass information directly, service calls to pass the data around, again, even in the most nominally functional microkernel design can be 100Xs the overhead.

Scaling up a microkernel to be as functional as Linux or Windows or OS-X grows exponentially. Here is from a person who is very PRO microkernel.

http://www.osnews.com/story/14353
And here we encounter the double-edged sword that is a microkernel; the easier a muK is to maintain on a local level, the harder it is to maintain on a global level.

That's why microkernels live in embedded and simple, ultra fast ultra reliable niches.

All the code is available for the entire kernel in Linux, you literally have every piece of code for every function available, so if it could be converted to microkernel architecture, it would have been already. I would suggest that duplicating the current 6M lines of kernel code into true microkernel architecture (keep in mind, "all" you have to do is subdivide it) would take a thousand coders 2 years, and be truly, globally, unmaintainable.

TripleII

Reply
Flag

0 Votes

+ -

Stil there should be an alternative.

Sagax- 24th Sep 2009

It seems most Linux is based upon 3 kernels. Those are Debian, Redhat and Slackware. From these arise a myriad of distributions. It seems to me that there ought to be at least one lean mean kernel available for developers and distribution compilers. For now though we must just cut back code on existing projects. (Example: AntiX is a lightweight reduction of Mepis. You probably know of several others)

0 Votes

+ -

Take a look at Haiku

cjc5447 22nd Sep 2009

If you are concerned about Linux kernel bloat, then take a look at Haiku, the open source successor to BeOS:
http://www.haiku-os.org
Fast, loads quickly, micro-kernel designed for responsiveness, multi-threaded, great performance.
After many years of hard work it's now an alpha release, they need developers to jump in and help.

Reply
Flag

0 Votes

+ -

BeOS

Linux User 147560 22nd Sep 2009

I really liked BeOS, wish it would have survived! Fast and stable... not many aps back in 99 but it was the fastest OS I had seen out there.

Reply
Flag

0 Votes

+ -

Haiku

Loverock Davidson 22nd Sep 2009

I wasn't a fan of BeOS when it came out but Haiku shows a lot of promise. I used it for a little bit in virtualbox, its pretty decent.

Reply
Flag

0 Votes

+ -

ummm

IssacS 22nd Sep 2009

I hate to be a stickler, but the last time that I checked Haiku was a Hybird Kernel, same as WinNT/XP/Vista/7 and The Mac OS X/10.x series.

I will admit that I have had an eye on Haiku for a while. I am rather interested in what ideas are forthcoming for R2, given how they have been working for years on R1 it would be nice to see if there are any ideas going forward.

Reply
Flag

0 Votes

+ -

Haiku is better then what some think...

Robert Wheeler 24th Sep 2009

This is the real deal here.. BeOS was (still is) one of the most fasiest, most stable os's out there, and Haiku fallows right in the same foot steps.. While there is alot more to Haiku, BeOS 5.0.4 was limited on more modern hardware. I had recently installed, and I use Haiku os right now as we speak, it's light, fast.. (booted on my quad core with 4gb ram) in under 4 seconds flat. You want to talk about fast. I Installed it, and it knew all my hardware.. even automaticly was able to use my connection.. just that easy. While I did foot with it to make it even more quicker, out of the box was most impressive. It put's even mac os x to shame in some areas. It's true, just like BeOS Haiku lacks apps. It's a good thing most BeOS stuff works like a charm under it. I am a developer anyway, so I would take notice of things. I think the team has done a wonderful job.. I can tell they do have work ahead, but never the less, it's impressive.
Lastly "given how they have been working for years on R1" My friend it's new.. really all new.. It's not BeOS with some code slaped on it. Building an OS does take years. The developers have worked hard without getting payed to bring Haiku where it's at. Give it 2 years and see where they will have it.

0 Votes

+ -

RE: Does it matter that

none none 22nd Sep 2009

When you DL a Linux Distro the kernel needs to be able to detect all kinds of hardware. On the other hand, if you were selling Linux Netbooks there would be no need to have kernel support for all kinds of hardware. You'd need to support only the hardware you're selling.

Or embedded.

That might not impact the developer experience but on cheap hardware it would improve the user experience.

happy

Reply
Flag

0 Votes

+ -

Bloat means "full of useless stuff" such as gas!

kd5auq 22nd Sep 2009

Any "universal" distro that runs on anything IS probably full of useless stuff "just in case it might be needed someday".
I wish it was easy (and automatic) for the "non-programmer" to compile a lean fast kernal on the fly for the hardware installation at hand.

Reply
Flag

0 Votes

+ -

Agreed- that would be pretty cool

unclefixer@... 23rd Sep 2009

And extremely useful!

Reply
Flag

0 Votes

+ -

RE: Does it matter that

Loverock Davidson 22nd Sep 2009

Yes it matters, it matters a great deal. Who would want to run linux in the first place? This bloatedness just gives them all the more reason not to run it. You admit it yourself that the developers won't develop on it, which means less people using it, which would make me happy. I can't see running linux at all, but especially not if its got duplicate code, duplicate applications, and who knows what else the developers installed into it. Just shows that linux lags behind the game by years if Apple and Microsoft are optimizing their code at this point while linux is still trying to add code.

Reply
Flag

0 Votes

+ -

Defense?

Viva la crank dodo 22nd Sep 2009

duplicate code?

And duplicate application that are part of the Linux kernel?

Reply
Flag

0 Votes

+ -

So if you don't like or use Linux...

djchandler 22nd Sep 2009

Why do you care so much about dis-crediting Linux? You seem awfully invested time-wise.

It does make me wonder as to your motives. Are you just being altruistic, steering the great unwashed away from the dreaded beast that is Linux?

And could you ramp up the melodrama please?

Reply
Flag

0 Votes

+ -

I considered that

Viva la crank dodo 23rd Sep 2009

but he would be willing to provide sources for his arguments or corroborating links for such masses if that was the case. Instead he only provides personal anonymous vague experience which, even if true, only proves that it did not work for him.

I have asked him to back up claims to but he refuses or is evasive by claiming he had provided evidence. (The only link he ever provided was to the front page of a Linux org site which supported nothing he said).

0 Votes

+ -

LD's problem

fatman65535 24th Sep 2009

I think LD's problem with Linux stems from an abortive attempt he made some time ago to compile a kernel for an RCA 1802. The compiler could not handle the strain.

Later on, I think he compiled a kernel for the x86 architecture, and expected it to run on a Motorola 68000. (Hell, they both have a "6" and an "8" in them!)

LD seems to have been stuck in the past. In fact, every time I see one of his posts, for some strange reason, the Jethro Tull song: "Living In The Past" plays in my head. In fact, it is playing in my head right now.

0 Votes

+ -

Message has been deleted.

IT_Guy_z Updated - 26th Sep 2009

Reply
Flagged

0 Votes

+ -

Message has been deleted.

djchandler Updated - 26th Sep 2009

Reply
Flagged

0 Votes

+ -

Message has been deleted.

stillgolfing Updated - 26th Sep 2009

Reply
Flagged

0 Votes

+ -

Message has been deleted.

fatman65535 Updated - 26th Sep 2009

0 Votes

+ -

If it is, I haven't noticed

eMJayy 22nd Sep 2009

The only thing that I've been noticing is that Ubuntu is only using around 220 - 240MB Ram at boot up, regardless of the amount of Ram in the machine. That's not bad for an OS that's using a compositing engine. Meanwhile, all my PCs that are running Windows are using 450 -600MB at boot up and that doesn't even include the stuff that Vista and 7 are caching to the hard drive. So if Linux is 'bloated', it's still got a long way to go before it gets to the state that Windows is currently in.

Reply
Flag

0 Votes

+ -

Run init 3 as root.

TripleII-21189418044173169409978279405827 22nd Sep 2009

You will get closer to what the kernel is using in terms of RAM. At least 120MBytes of the above usage is the Gnome desktop which is not part of the kernel.

You can get back to desktop if you run init 5 as root again, or just reboot with init 6 (I remember a long long time ago Gnome didn't like being started manually with init 3 if it was already running).

TripleII

Reply
Flag

0 Votes

+ -

Same here.

djchandler 22nd Sep 2009

Considering that you can get reasonable performance in 512MB on an Atom CPU, I'd say the bloat isn't that bad. Windows 7 requires 1GB of ram.

If I'm not hampered performance wise, getting better benchmarks on the same machine than with Windows 7, I'm pretty happy with Linux.

How does OS X stack up versus Linux in terms of ram usage and benchmarks?

Reply
Flag

0 Votes

+ -

Nobody runs all the kernel code.

TripleII-21189418044173169409978279405827 22nd Sep 2009

As part of the Mandriva install, it asks you if you want to remove all the modules you won't need. So if you take the entire kernel (which is the MS model, disabling various pieces depending on the version you purchase), you can use as much or as little of the available kernel code you want to use. We are developing a new product using CentOS and literally, every unused module or hardware driver or function we will not use in the product has been removed. This is more for security, not speed since unused modules don't use resources (OK, some RAM for the anal, lol). As a perk, it takes about 5 seconds for the service modules to start loading after the kernel on boot.

So I don't see it as being a problem, lots of hooks to modules you can choose not to use. As already said though, even using a "bloated" kernel, things are pretty stable.

TripleII

Reply
Flag

0 Votes

+ -

Informative.

djchandler 22nd Sep 2009

Agreed about the "blobs" and what not included in the kernel. A lot of it is hardware drivers.

I've heard nothing but good stuff about CentOS. But what in particular makes it your choice for your project?

Reply
Flag

0 Votes

+ -

Cost and based on Redhat Enterprise.

TripleII-21189418044173169409978279405827 22nd Sep 2009

Redhat has a lot of name recognition and while we use it,we are experimenting with CentOS (well, more than an experiment now) because it is based on Redhat and customers are already familiar with procedures on Redhat. Since we are semi-intelligent here too, we will maintain the OS as opposed to Redhat. Since it is binary compatible with RHEL customers that insist on paying for their support (which some do, they want the guarantee) we are fine with our development.

TripleII

0 Votes

+ -

Drama

djchandler Updated - 22nd Sep 2009

You have to motivate people. Linus may be guilty of a little hyperbole.

Reply
Flag

0 Votes

+ -

RE: Does it matter that

DemonX 22nd Sep 2009

Didn't anyone read his full actual comments? He said it couldn't be helped. Its a reflection of the number of included features and drivers. Its why complaining of bloat in windows is so stupid. If you want all the features and included hardware support in windows or linux then stop complaining. And If you want your linux slim and trim then shut up and roll your own. The only time I've ever compiled the kernel was to include additional hardware not to remove hardware. Being "bloated" just means its more likely to support what you want to run it on.

Reply
Flag

0 Votes

+ -

RE: Does it matter that

IssacS 22nd Sep 2009

Well given the way that linux use to operate, I think this kind of talk that makes one question when should a 2.7 (developing-branch) should be called to clear up the code base.

As much as new features are coming to the code, I think (and it would seem that Linus either see this and/or agrees with me) that alot of the additions has been rather hasty particularly as of recently. Now give that Linux has yet to brake under the strain, it could be a while before we see a development slowdown and a call to 2.7/2.8... but if Linus has noticed the bloat, I would hope that it's being considered.

Reply
Flag

0 Votes

+ -

Bloat ALWAYS Matters. But Compromise is ALWAYS Necessary.

trentreviso 22nd Sep 2009

Good software design maximizes the use of whatever hardware is available. The fastest and most efficient programs (by far) are written in machine language. So, end users would benefit from enormously improved performance if all modern OSs were coded in x86 (or whatever) machine language.

But, that would remove the portability of the code across processor platforms. It would slow down dramatically the patching of vulnerabilities and the addition of new features to the software. It would increase the number of man hours necessary to finish a project, and would raise the cost of development substantially. It is not practical.

So, some compromise between bloat and ease of development always needs to be struck. What Torvalds is saying is that Linux coders need constantly to struggle not to let the bloat get out of hand.

That Linux has done this well to date is evident. Ubuntu ships on a 700 MB disk and is very nearly the functional equivalent of OS X and Vista, which are four times Ubuntu's size. And Ubuntu includes a full suite of office applications, media players, utilities, games, and other software on that disk. Relatively speaking, Linux is a model of software coding efficiency.

But that is not to say that Linux is not bloated. It has swollen to many times a size that would seem to be absolutely necessary and will continue to bloat as it develops. The challenge will be to keep the bloat from escalating out of all reasonable control, as Microsoft experienced with Vista.

That is what Torvalds cautions us against.

Reply
Flag

0 Votes

+ -

RE: Does it matter that

frj111@... 23rd Sep 2009

I started playing with Linux again when I installed Ubuntu 8.04 last year. Prior to Ubuntu I had used the Mandrake version of Linux in several flavors including one DOS 6.22 version. Obviously a Core 2 Duo system is considerably more advanced than an Intel 386. So one would hope that the kernel was a tad more advanced.

Is it still in fashion to say ?It is not your father?s Oldsmobile?? Or will people look at me and ask what is an Oldsmobile? The times they are a changing.

Reply
Flag

0 Votes

+ -

RE: Does it matter that

dcolbert@... 23rd Sep 2009

"Code bloat is unimportant, as soon as code bloat affects Linux".

I've been pointing out how bloated an install of Ubuntu is for awhile now. It is easier to install XP on a 4gb Eee PC than Ubuntu, with more space left over once you're done on the XP system.

As soon as Linux becomes guilty of what it has been accusing the competition of for years (usually in an attempt to mimic features that the competition has *offered* for years), suddenly that point is "unimportant" or "not the same thing" as it was when it only affected that competition.

The Linux commmunity has *always* wanted to have their cake and eat it too.

Face it, by the time Linux is ready for the desktop, it will be virtually indistinguishable from OS X or Win32/64. Which is fine... in which case, choose what you like, and what works for you, and stop making it a holy war.

Reply
Flag

0 Votes

+ -

It's like past/present/future dollars

jeverettk 23rd Sep 2009

To be fair, you've got to do a net present value calculation:

Bloat in 1991/processing X memory 1991 VS. Bloat today/over current processing & mem.

Torvalds may feel a twinge of nostalgia, but I think he's mainly just mouthing off.

Reply
Flag

0 Votes

+ -

Reason it does matter...

Spiritusindomit@... 23rd Sep 2009

Is because it always matters when people bash windows. Case of having your cake and eating it too.

Reply
Flag

0 Votes

+ -

Another blog on ZDNet on this same topic...

Filker0_z 23rd Sep 2009

Under a previously posted article (http://blogs.zdnet.com/BTL/?p=24643) by Sam Diaz, I posted the following response under the title "Linus has a point"

True, the Linux kernel is not what Linus wants it to be. There are too many layers, and too much code is executed between the application and the hardware. Does this make it a bad system? No, it means that an effort should be made to look at what the Linux kernel is now, and learn from it to develop a new kernel with the same user-space API that is leaner, more efficient, and takes fewer lines of code to implement.

Unlike MS Windows, the UI for Unix is not in the kernel, it's a user-space application. We can't compare the two directly, so I won't even try.

Linux is too monolithic; most device drivers should be moved out to a class of "user" processes that each execute in their own address space. File systems, too, should be implemented in user-space, and register themselves with the kernel as they are loaded. Several other parts that are now built in to the OS kernel should become processes outside of kernel space. Making this separation is what makes a kernel into a microkernel.

That's the theory. As Linux is, it's still a good kernel, but it's based on an older model. The codebase has grown to a point where it's not as much fun to work with; too many interdependencies, where a change may break things. That's the scary part.

What's needed in a Linux 3.x development effort, creating the lean, efficient, and small microkernel, a driver process model, a service process model (for filesystems, networking, etc.), and the like. The goal should be to provide a compatibility interface for userland apps built for 2.6 apps. A native 3.x interface should be modular, and it should be possible to load multiple userland ABIs at the same time. Ideally, drivers, services, and native apps could share the same ABI.

This is a forward looking view. Linux, as it now stands, is vibrant and growing, and I would not take Linus's comments as a rejection of Linux; he just thinks it needs to take a different direction than 2.6 is headed.

Linux has not failed; it just can be better.

Reply
Flag

0 Votes

+ -

Well...

IssacS 23rd Sep 2009

I sort of agree with you, but what you are suggesting is a total rejig of the kernel architecture and how the system functions currently. I am find that a little drastic, since the system (even under this strain) hasn't have a massive catastrophic failure for any release... but that does make the call for a 2.7/2.8 louder, at least to clean up code and clear out a lot of cruft.

I will admit that if Linux is to take a restructuring, mircokernel is the least likely thing to be done... they are highly complex things, and it could take years along with massive resources to get something like that off the ground. I am not sure that all companies that have invested with Linux would be willing to take that downtime to get development up to par.
I would say that a move to something more like a hybird kernel would be likely (like Mac OS 10.x and WinNT), but Linus (if my memory serves) doesn't have much love for the concept of that type of kernel... it's questionable if he would consider it now. With that said, however, farther modulation of system functions would be welcome to the codebase... drivers in particular.

Reply
Flag

0 Votes

+ -

RE: Does it matter that

jurgen.manycolored@... 23rd Sep 2009

You can recompile the kernel and cut it way down. WAY down. And you can get a distribution where this is partially done. So what's the problem??

Reply
Flag

0 Votes

+ -

GNU/Linux should take a break.

Bilmekanikeren 23rd Sep 2009

In the last few iterations of the kernel as well as some distributions there has been a fairly large amount of bugs.
Such as: Sound works on one kernel, but not another. Ditto for wireless cards. I think developers should stop developing features for a while, then clean up and optimize what they already have.

Reply
Flag

0 Votes

+ -

RE: Does it matter that

discar 23rd Sep 2009

It matters that the kernel is bloated. Linux is great in that
basically, the kernel is basically identical between what you
put on that 16GB 4-core workstation sitting on your desk,
and what you put on that 1GB notebook, or that 512MB
netbook, or that PDA (cross compiling as needed, of
course).

But as the kernel gets more bloated, the less likely you'll
see low-iron platforms using it: if a developer needs to
weed down the linux kernel, why not just weed down the
Windows Mobile kernel?

Reply
Flag

0 Votes

+ -

RE: Does it matter that

valvestate@... 23rd Sep 2009

Thank you for (sort of) pointing out that today's systems are much more capable of handling bloat than the systems Linus Torvald used when he had his vision of how efficient a kernel could be. A previous ZDNet article weighed in on this subject referencing a CNET article that was later appended with a key factor about the difference between Linux and Windows. Linux is scalable, as in as bloated as you want it to be. A person who knows what they are doing can install just the parts of Linux they want to run the apps that they want. As other commenters to the previous ZDNet post pointed out: Being all things to all people isn't Linuxs' forte but there is likely to be a flavor of Linux that will suit a particular need.

The only analogous of this I know of in Windows other than illegally hacking the Windows source code, is to use a program called nLite which allows a user to select or deselect components of the Windows OS to create a bootable ISO of a parred down Windows OS. This venture is far from the same as selecting components at the time install of the OS since an ISO has to be created on a drive, components selected/deselected and another ISO has to be created and that ISO has to be burned as a bootable disk. This process points up why Windows would more properly be referred to as a "client" rather than an OS. Removing the apps integrated into Windows like Outlook, Windows Media Player or the dreaded Internet Explorer reveals that parts of the OS are integrated into these apps. This is what I would term hyper-bloat. A practice that would be the extreme opposite of Mr. Torvald's vision. If you want to play around with nLite for rolling up your own Windows install, unless you have a corporate copy, have fun jumping through the hoops Microsoft has set up. And lastly (about nLite) if you remove IE you can't update the OS. There are other ways to update but Microsoft has to make sure that anyone using Windows is stuck with it's rotten, non-compliant browser.

Reply
Flag

0 Votes

+ -

RE: Does it matter that

pool7 23rd Sep 2009

Well, at least you can choose which modules the kernel has/uses when compiling, so you can still make it less bloated... I don't think we'll ever see something in the likes for Windows.