thelinuxlink.net

Debian is now producing a multiple platform network boot cd. One cd has 64bit/x86/ppc. This is a real plus if you have to support installing multiple platforms. We do not have 64 bit yet, but we do have x86-32bit and ppc's.

http://cdimage.debian.org/cdimage/daily ... ch/iso-cd/

That's very interesting, the only thing I wish Debian could do better was support they releases much longer. I mean it take like 24 months per release and there only supported for 24 months or so. Centos is like 7 years.

allix wrote:That's very interesting, the only thing I wish Debian could do better was support they releases much longer. I mean it take like 24 months per release and there only supported for 24 months or so. Centos is like 7 years.

It's generally one year after the release of a new stable distribution, so...

Potato was supported 3 years
Woody was supported 4 years
Sarge was supported 3 years
Etch should be 2/3 years.

But who uses stable, anyway?

Wally Balljacker wrote:
But who uses stable, anyway?

Troels

jokes aside, organisations might want to use Debian.

allix wrote:

Wally Balljacker wrote:
But who uses stable, anyway?

Troels

If you're referring to CentOS, then you would be right, I do use that in a few cases, although on my desktop I use Fedora.

Tsuroerusu wrote:

allix wrote:

Wally Balljacker wrote:
But who uses stable, anyway?

Troels

If you're referring to CentOS, then you would be right, I do use that in a few cases, although on my desktop I use Fedora.

That is until they quit supporting Fedora 8. What's a Troels to do?

With debian and the like distros you can easily upgrade when a new version comes out, so virtually you have unlimited support so far. The other distros are little more involved to do that.

Wally Balljacker wrote:That is until they quit supporting Fedora 8. What's a Troels to do?

There'll be a few options:

1. Take another look at KDE 4.x at that point in time (Will probably be around KDE 4.2), and see if it's usable for me.
2. Continue to run Fedora 8 (Nobody forces me to upgrade) until KDE 4.3 comes out.
3. Move to another distro temporarily.

eddie wrote:With debian and the like distros you can easily upgrade when a new version comes out, so virtually you have unlimited support so far. The other distros are little more involved to do that.

That's not an option for businesses that need their systems to just run and continue to run for a long time. Time is money and even a little downtime can be a lot of money.

With pre-prepared images and iscsi (or other remote boot method) where you can have diskless servers, an os change is just a reboot of the server. If you have the machines mirrored, there is virtually no down time. I am sure the same thing can be set up with virtual machines also. Even on a small scale, if you have systems properly partitioned and images properly pre-prepared, an os change is not that much of a life change.
We used to multicast images to student labs after the gaming club at the college used their own images on student machines over the weekend for a convention. Student machines were up and running first thing Monday morning. All student data was server based or on students personal removable devices. I will say that IT management was not thrilled in have to do this.

Tsuroerusu wrote:

eddie wrote:With debian and the like distros you can easily upgrade when a new version comes out, so virtually you have unlimited support so far. The other distros are little more involved to do that.

That's not an option for businesses that need their systems to just run and continue to run for a long time. Time is money and even a little downtime can be a lot of money.

I'm not following you. Firstoff, if a business has no tolerance for outages, even for a few minutes during a maintenance window, they had better have a much more bulletproof solution than a single COTS x86 box running linux, whatever flavor it is. Other than that, what do you mean here?

Snarkout wrote:

Tsuroerusu wrote:

eddie wrote:With debian and the like distros you can easily upgrade when a new version comes out, so virtually you have unlimited support so far. The other distros are little more involved to do that.

That's not an option for businesses that need their systems to just run and continue to run for a long time. Time is money and even a little downtime can be a lot of money.

I'm not following you. Firstoff, if a business has no tolerance for outages, even for a few minutes during a maintenance window, they had better have a much more bulletproof solution than a single COTS x86 box running linux, whatever flavor it is. Other than that, what do you mean here?

Well, what I meant is that the impression I get, is that business don't like upgrading their machines every other year. The longest Debian life-cycle so far seems to have been around four years. Well, Red Hat's still up there at like 7 years, I'm sure Novell does something very similar.

Yeah - I believe that's true. We still have some oldass RHEL 2 boxen kicking around on our network, and we still get occasional security updates. Believe me, if I had the option of doing an apt-get dist-upgrade on them, I'd be happy as hell - trying to get modern hardware to work on 2.4 kernel is about as fun as driving a rusty spike through your nutsack. Of course, if the dumbfuck who did the installs years ago had partitioned the disks so I could reinstall the OS non-destructively, this would be more or less a non-issue.

eddie wrote:With pre-prepared images and iscsi (or other remote boot method) where you can have diskless servers, an os change is just a reboot of the server. If you have the machines mirrored, there is virtually no down time. I am sure the same thing can be set up with virtual machines also. Even on a small scale, if you have systems properly partitioned and images properly pre-prepared, an os change is not that much of a life change.
We used to multicast images to student labs after the gaming club at the college used their own images on student machines over the weekend for a convention. Student machines were up and running first thing Monday morning. All student data was server based or on students personal removable devices. I will say that IT management was not thrilled in have to do this.

That's a very good point , Xen and virtualbox , vmware has been promoting that use for a long time .
Out of interest was the gaming club windows based? if so, well windows xp has almost 20 years of support.

Let me explain. A mirror or cluster is where two or machines are doing the exact same thing. This aids in sharing the work load and if one system goes down there the system still runs, but with fewer systems. Since it is a mirror, you can add and remove machines without a work stoppage per se like a raid system for hard drives. You could take one or more machines at a time an do what's needed and return it to the mirror/cluster. If you are running iscsi via gpxe, your servers could be set up to run diskless. Which makes the os change even easier. Even in a small business you would want to have an extra duplicate machine (redundancy) just to prepare images or updates so that a cable swap to the updated machine is all you need to do. When the original machine is updated, the network connection can be restored. Using virtual machines makes it even easier. That is one of the reasons I like ltsp so much. I do not have to update any clients. I just update the server.