As dominant systems and software companies continue to pour millions into scaling Linux for the enterprise, some observers now believe the open-source OS’s future is to serve as a distributed, network-based operating system. Read the article at InfoWorld.
Linux Future Scales Out
About The Author
Eugenia Loli
Ex-programmer, ex-editor in chief at OSNews.com, now a visual artist/filmmaker.
Follow me on Twitter @EugeniaLoli
There has been a huge initiative to make linux into a desktop usable system, and now they are trying to make it into a widely scalable system for corporations to use too. While they happen to be doing a good job of getting both to play nicely right now I can’t help but wonder what this brings for the future. Eventually the kernel is going to get bloated beyond all belief supporting (anybody know how much the codebase grows with each release? that would be an interesting statistic) all of this plus, maybe even cause speed problems because of the optemizations necessary to run on MP systems versus UP systems (hell, we might even might want to ignite some microkernel versus monolithic fights for fun, wait no those are violent :-P).
If linux wants to be both an os like mentioned in the article above and a desktop os, you are going to see some large disto changes. There will be specific server (and mp) linux distros and specific desktop distros and the differences between distros will be more than just what is packaged with it and what little tweaks they have. The thing is, not everybody likes to compile their own kernel (certainly not your average desktop user) and yet they would still like to have a responsive system (and a bloated kernel doesn’t help that) so the distros will come with custom kernels already in place.
By the early 3.x versions of the kernel, we may actually see two different code bases, one for systems withh large amounts of processors, and one for desktop. Like I said, I hope they play nice.
The Linux Kernel is totally modular (did you seen that µLinux have been even integrated into the 2.5 tree?), so you can replace some part of it with another (sometimes you have to patch it) ans it’s the job of the distrib to offer facilities to do that. And if you have never seen, current distribs (Mandrake, Gentoo or Debian, etc) already offer several kernels, for different archs, or for different configurations. I don’t think that’s a problem.
I think that Linux will lose mindshare if it doesn’t aggressively pursue the desktop. Even if it never catches up to Windows in that arena, I don’t believe that being a server-only or network-only is in its best interest – Netware, for example.
But the abundance of distros has reached a point where it’s counterproductive. Distrowatch.com now lists 94 distributions – I mean, really. More effort into the areas where it’s lacking and less into the distro-of-the-month
Regarding the 94 distro, the other 95 don’t matter one iota in the grand scheme of things. In around 6 to 12 months, there will be two camps, UnitedLinux and Redhat, everything else won’t matter one iota.
People who want to feel “l33t” can still continue to use the frindge linux distro’s, whilst the majority will stick to the mainstream versions that will be available.
As for applications, there needs to be commercial development and personally, IBM might as well burnt the $1billion because there has been nothing added to Linux in the 2years they have been on the bandwagon. If they wanted to do something, they could have ATLEAST got Lotus to port Notes and Smart Suite to Linux natively, or atleast helped wine to get Lotus Notes and SmartSuite to work.
While I don’t exactly agree with you when you say IBM has done nothing for Linux in the past few years, they have ported it to all of their systems.
Unfortunately, they are still using AIX on their new Power4 workstations. Really, the only thing that needs to be done is make a driver for the graphics card and then Linux would be a fine alternative, probably superior in some respects.
On the x86 side of things, IBM should buy Netraverse so they can use Win4Lin. If IBM were to open source Win4Lin, then they could easily integrate it into a desktop package where users demanded a “High level of windows compatibility.” This would of course require Microsoft getting money from the sale of WindowsME, but at least people would be using Linux more. Other users who didn’t care as much could just get a Wine/WineX/Codeweavers enhanced dekstop. Eventually IBM could help combine these different packages and WindowsME would no longer be needed at all.
This would allow IBM to run Windows apps on their Power computers as well; all they would need to do is use something like http://www.transitive.com/ in conjuction with their Wine(X)/Win4Lin package, and they would be set. Sure, it is no small programming project, but for IBM, it would be quite a small task to accomplish.
I agree that IBM should port its Lotus products, I know a corporation that would like to switch many of its systems to Linux if only SmartSuite were working in Linux w/o the hassles of dealing with Win4Lin.
Hey if IBM Believes so much in open source why does it not open source OS/2 and also if it believes so much in linux why no SmartSuite for Linux.
Well the game plan for IBM & the so called commercial backer’s of OpenSource is to milk the open source movement for their own benefits.
Sad but true 🙁
version size
0.01 71k
1.0 1.2M
1.2.13 2.2M
2.0.39 7.2M
2.2.23 18.8M
2.4.20 32.4M
2.5.50(2.6.0?) 37.1M
but the size of the core kernel code has been growing much more slowly. most of the added size you see is support for different architectures and devices which bloat the size of your kernel and you dont need.
i always thought it would be a great idea if someone patched the kernel to rsync only the files it was about to compile, so dropping a large driver would speed up your download.
however, the speed of internet connections has kept pace with the kernel, meaning it _still_ takes about 5 minutes to download, as it did at ~1.2.0
If 2.4.x is 32MB then how come you can save it to a floppy to test?
the SOURCE CODE is 32MB…the actual binary is usually much smaller than that…plus the binary kernel image is usually compressed so that it will definitely fit onto a floppy disk
-bytes256
if you compile your 2.4.x kernel with
(make menuconfig)
make bzImage
it’s about 550 to 750 kB depending on what you configured.
😉
Matthew….. Regarding the 94 distro, the other 95 don’t matter one iota in the grand scheme of things. In around 6 to 12 months, there will be two camps, UnitedLinux and Redhat, everything else won’t matter one iota.
So Debian will just disappear?
I doubt we’ll be seeing the distro landscape change significantly anytime soon. The fact that there are 95 distros on distrowatch isn’t really relevant, only perhaps 6 of them really matter.
As for applications, there needs to be commercial development and personally, IBM might as well burnt the $1billion because there has been nothing added to Linux in the 2years they have been on the bandwagon. If they wanted to do something, they could have ATLEAST got Lotus to port Notes and Smart Suite to Linux natively, or atleast helped wine to get Lotus Notes and SmartSuite to work.
Sorry, this is just plain wrong. IBM have made back their $1 billion already in server sales and consulting apparently (I don’t recall where I read that but it was news article in a reputable publication). Porting desktop applications isn’t what they’re interested in (at the moment), it’s totally server side stuff mixed with consulting.
And FYI Lotus Notes runs just fine on Wine:
http://users.theshell.com/~vinn/ss/Thomas_Wickline/2002_11_30_01212…
Also that’s actually 34mbs tar. It’s around 100mbs (less or more, i’ve forgoten) when it’s untarred, but yeah that’s just source.
Here is where Linux stands. There are 5 main areas of computing
1) Enterprise very large systems.
Linux is quickly becoming a major player in super computing with cheap clustering. As of last year the fastest computing machine (a 64 way Itanium 2 system from SGI) ran Linux.
In terms of mainframe type apps pretty much nothing. Solaris is making a huge play here which should lead the way for Linux.
2) Server
Everyone knows of Linux’s strength in this arena. In particular it has killed off most of the weaker Unux offerings. Microsoft server OSes are still strong but Linux is already putting a great deal of price pressure on Microsoft.
3) Corporate Desktop
In the last few years foreign nations have begun real experiments with moving towards Linux desktop setups. The problem is of course that 3rd office workers are better educated and smarter in general than their American counterparts so it is hard to know how realistic this setup is for the American market. With Europe, particularly Germany starting to utilize Linux in this way we should see movement in American corporations within the next 3 years.
Also see thin clients in the embedded systems section
4) Home / small business
Linux may fall further behind in this market. Microsoft has publically shifted their focus of R&D from primarily corporate desktops to primarily digital entertainment. Apple meanwhile is building a very strong offering in the home / small business catagory both on the server and the destkop front.
However the rise of the under $500 PC will help Linux tremendously. If a whole system (monitor included) can be purchased for under $200 we might see a return the habits from the 1980s (commodore 64…) except in larger numbers. Its hard to know if this plays out or not.
On the other plus side the work on the corporate desktop will help for the home / small business market. Linux needs to improve the productive apps usability for this group of customers.
5) Embedded systems
Linux continues to make inroads here. Microsoft CE’s refusal to support X and to include many terminal emulations is killing it in the thin client arena; and Linux systems are by far the better buy. Whether thin client computing makes huge inroads in corporate America remains to be seen.
In the handheld market Hancom Suite has put Linux is play. Palm is clearly starting to slip but with small cell phones being able to run Palm they might be getting a second wind.
In the comercial embedded systems market (no user interface) Linux is becoming the OS of choice though this market remains very competitive.
I don’t really care for the Wine-based apps for the long term. They should only be a crutch to be used when a native port CAN”T be gotten – say, Internet Explorer.
If Big Blue believes in Linux on the desktop, then it behooves them to port SmartSuite or, donate enough of a codebase to build on.
Hey, Sun gave away the code to StarOffice ( and some developer time, if I’m not mistaken).
How many of those 95 distros are commercial distros and how many of those that are commercial distro’s are aiming at the home desktop full steam ahead, ie eliminate the server, business desktop, hobbiest distro etc.. Make a guess please ! I would say that maybe there about 3-5 commercial distros who are aiming squarely at the home desktop as their core audience. The rest you wouldn’t find at your local CompUSA or at least would not be advertising for home users but experince users looking for a nice server OS to use or business desktop .
yeah, sorry, that was source.tar.gz size. the binary size doesnt change very much, on x86 is usually around 300k for me.
the gzipped source is a good measure of how much ‘information’ is in a distribution, though.