Slashdot Mirror
Constructing a Windows-Less Office
joewakeup writes "This article at CRN analyses why today is the best time to consider building a pure Linux information system, from servers to... desktop. Among all the arguments, one of the arguments is the low cost of Linux offerings compared to Windows based-solutions. Worth a read."
All but one of the servers they were using ran Linux (the remaining two were running Solaris and NT for software requirements). I worked under the network admin, and during the whole time I was there we never even had a glitch with the network.
All of the engineers were using Linux on their desktops and it worked beautifully. The remaining desktops were running Win98 for the HR, marketing and finance groups because the software they were using required it.
It's not quite the Windows-less office that the article was discussing, but it was pretty close. I've seen the wonders of the Linux-based network and I like it.
There is no escape from The Muffin.
Sorry, but if I added up all my time spent fixing broken Windows, and compared it with the cheerful hacking I do on Linux or BSD, Windows would come out far more expensive.
Of course, YMMV, but in our 300-plus node network of Windows boxes, you can always guarantee one thing: they break when you need them most.
Personal computer systems are brittle as hell, and, as far as I'm concerned, running Windows is no guarantee that your day won't be wasted. Perhaps NT decided to blue screen because there wasn't a PS/2 mouse plugged in (true story).
My personal obsvervations indicate to me that it is a fallacy that Windows is easier to maintain. Tell that to our IT guys.
-- clvrmnky
it's about twice as slow as Win2K/XP on the same hardware.
That really depends on what you're running.
I'll probably never hear the end of it if I say this, but I'm going to say it anyway: The most popular window managers for the XWS are also the most bloated.
Have you ever used Nautilus? It is a very pretty interface, but it is slow as all hell on a machine of reasonable specs. (PIII 500 / 256MB) Now take Gnome and Nautilus, plop it on to a system, and yeah... it's not going to perform as well as it should. Granted, the XWS isn't the best performing GUI out there, but the 4.x rewrites are solving a lot of those problems.
I've used Gnome/Nautilus as an example above, because I know less about the newer KDE releases with regard to frendliness, performance, and bloat. If someone would be kind enough to fill me in on how KDE is in these respects, I'd appreciate it.
Anyway. Gnome is a pretty hefty download, and tries to shove all of the crap they think you'll need into the package.
If you set your users up with something like AfterStep (which, by the way, can fit on a floppy), ditch the desktop pager, show them how to use Wharf and the Winlist, and install the apps they will need. Configure Wharf to make it easy to get the apps, then smack everything onto a kickstart server or something. Then whenever a new box enters the office, just kickstart the image on to the box and there you go. No configuring, and it would make administration much easier. (You could probably also hack in some cronjobs on the server and the workstations to automatically keep all packages up to date, but that's beyond the scope of this comment.)
This way, they have a fast, clean window environment, the apps they need, and the benefits of Linux.
-- Give him Head? Be a Beacon? :P)
(If you can't figure out how to E-Mail me, Don't.
is getting it to work with all the hardware. by the time i got all my stuff configured on my own systems, my head was swimming. hunting down drivers on google and editing config files by hand was very educational, but not particularly speedy. when i imagine doing the same for every machine in even a small organization, my head wants to explode.
of course you could just buy a machine with linux pre-installed, but then you get the choice of a dell latitude model X, or dell latitude model X. and installing linux on a machine that came with windows on it rather mitigates the lower cost argument, since you've already paid for the windows license. or you could buy individual components that have linux support and form a santa's workshop to assemble machines. again, not particularly cheap or speedy.
so, it's not the lack of windows app alternatives that's holding linux back in the workplace, because staroffice, gimp, etc., cover 99% of what your average user would need to do. it's also not the vaunted inertia that everyone makes a big deal out of, because the interfaces for open source alternatives almost completely mimic their windows cousins. believe me, the learning curve is no higher for telling people how to use the OSS version of a spreadsheet program than the windows version itself.
imho, once it's as easy to get linux running on a given machine as it is windows, the major obstacle to moving your business platform from windows to linux will be gone. until then, all the security, stability, and financial arguments in the world are not going to outweigh the perceived headache of having all your tech staff running around for years trying to get the workstations config'd properly.
Doesnt' Brick Level Backup destroy single instance storage? Meaning, I might have a 100GB .mbd file, but when I Brick Back it up, that might grow to 1000GB because all those single instance messages will get individually saved into each mailbox.
Backing up 1000GB could take a long time. Brick Level Backup is considered a Bad Thing by the gurus of the microsoft.public.exchange.admin list.
You are, however, 100% correct about Deleted Item Retention.
Sure, but that's the Catch-22 Fucky Badger (now there's a nick I wasn't expecting to see with a +5) was talking about. My 1200 MHz Athlon flies with WindowMaker and gcc running in a wterm. But that's giving up what's making Linux an alternative to Windows in the eyes of writers like these.
I've used Gnome/Nautilus as an example above, because I know less about the newer KDE releases with regard to frendliness, performance, and bloat. If someone would be kind enough to fill me in on how KDE is in these respects, I'd appreciate it.
The newer releases are definitely getting faster but you still need pretty recent hardware to get snappy performance. And it seems like because of the kdeinit hack, starting apps is considerably slower if you're not using the KDE desktop than if you are. Still, the load KDE imposes comes from the KDE 2.0 architecture. There's no major source of additional bloat in the near future and it's pretty clear how to make everything faster -- drastically improve the way a GNU system loads C++ apps.
At any rate, the best way to get Linux going faster is to buy new RAM. I understand why people don't run and out for new processors or hard drives but RAM is so cheap now $20-30 will liberate you from the misery of hitting your swap.
What I'm listening to now on Pandora...
-
- On W2K to make your GUI more responsive, change the environment to favor "Foreground" apps instead of "Background" apps.
- On linux, renice your X to make foreground apps higher priority. Here's the command I put in my
.xinitrc:sudo renice -10 -p `ps -aux | grep "[^grep].*X \(vt\|:0\)" | awk '{print $2;}'`
-
- On W2K, turn off services that you don't need
- Same thing for linux, lots of services get started -- but do you need them running all the time? (e.g. if you're not developing for your db app that day, turn off the DB server)
- For some reason, at least with RedHat, all of the apps and libraries are installed with debugging symbols. Who knows what their reasoning is, but that takes apps longer to load and slower to run. Have you ever run a "Checked Build" NT or W2K that includes all debugging symbols? The same thing happens. What I do is boot to my rescue CD and strip every file on a regular basis (which also frees a ton of HD space). Be careful with this because you have to run ldconfig from the rescue CD before booting back to your default -- and it requires using params to ldconfig to point to the right mounted directories and spitting the ld.so.cache to the right spot. If you do it wrong, you will not be able to boot except to rescue CD.
- Compile a new Kernel (preferably with the new VM) and take out all of the unneccesary items and compile as much as possible into the kernel instead of modules.
I'm sure there are a hundred others, but these seem to do the most for me with least effort.Brick level backup doesn't destroy single instance storage until you restore something. Then a new instance is created for the restored items.
Both OpenOffice Impress and kpresenter are stable mature applications that can do most things a rational person would ever expect in a presentation.
So if you write
" There's no open-source software replacement for PowerPoint."
you are right. There is not one, there are TWO GPL apps to replace powerpoint.
Now if I look at the fact that SVG is a vector format (not a presentation format) and the fact that openoffice641 opens all ridiculous powerpoint stuff I get mailed by people, I think you should look harder before you propoese new projects to other people.
See http://www.openoffice.org and http://www.koffice.org for the apps.
Moritz
I work in an electronic music studio. I'd love to use Linux, but the apps just aren't there.
The fact that there's almost no development community addressing this potentially enormous market amazes me to no end.
On the linux-audio-dev mailing list, many things are discussed and software developed such as Ardour, digital audio workstation software for Linux, JACK (JACK Audio Connection Kit), a low-latency infrastructure for connecting audio applications, and several wave editors. Dave Phillips maintains a list of Linux sound applications--many are not that advanced but some are.
Work in this area is progressing, and many smart people are involved. In particular, Paul Barton-Davis, author of Ardour and the main force behind JACK, seems to be pursuing commercial possibilities of selling linux-based sound workstations under a company named Linux Audio Systems. You can read Paul's slashdot comments to see some of his opinions on the situation of Linux audio.