| |||||||||||||
|
| |||||||||||||
CentOS 6 making its way out to mirrorsCentOS 6 making its way out to mirrorsPosted Jul 9, 2011 16:54 UTC (Sat) by ewan (subscriber, #5533)In reply to: CentOS 6 making its way out to mirrors by dan_linder Parent article: CentOS 6 making its way out to mirrors
It should have done. Scientific Linux does what it does, and it does it better.
(Log in to post comments)
CentOS 6 making its way out to mirrors Posted Jul 9, 2011 21:02 UTC (Sat) by robert_s (subscriber, #42402) [Link]
"It should have done. Scientific Linux does what it does, and it does it better."
What, exist to serve the needs of a relatively small number of people and be around for about five minutes before becoming a stagnant backwater?
I realize that's a bit trollish, but I really don't understand people who put their horses on (and supposedly roll out on hundreds of machines running) distributions that have so little guarantee of being around and well maintained in 12 months time.
CentOS 6 making its way out to mirrors Posted Jul 9, 2011 21:08 UTC (Sat) by prometheanfire (subscriber, #65683) [Link]
Scientific Linux has been around since before CentOS (by only 4 days). I wouldn't call CERN small.
CentOS 6 making its way out to mirrors Posted Jul 10, 2011 9:49 UTC (Sun) by robert_s (subscriber, #42402) [Link]
CERN itself may not be small, but academic budgets are always notoriously threatened. All it takes is for the project to be squeezed out of the budget one year and things are in trouble.
CentOS 6 making its way out to mirrors Posted Jul 11, 2011 8:32 UTC (Mon) by pcampe (subscriber, #28223) [Link]
People are paid to work on Scientific Linux: this explains the extremely good schedule, and the support when it comes to pushing errata, as developers are not allowed to take vacation on the same time; CentOS has, at best, an opaque organizational structure, and some months ago they suffered some organizational difficulties as the owner of the CentOS web domain was out of reach.
It's possible that there are better ways to build a rebuild of RHEL, and that this two projects could somehow merge and reduce the overlap they have; but you can't say that, comparing to CentOS, SL has some kind of budget problem, it's exactly the opposite.
CentOS 6 making its way out to mirrors Posted Jul 10, 2011 9:55 UTC (Sun) by robert_s (subscriber, #42402) [Link] And while I'm at it, yes, CERN (even including the extended scientific community) is a very small market when compared to the world at large.
CentOS 6 making its way out to mirrors Posted Jul 11, 2011 1:28 UTC (Mon) by ewan (subscriber, #5533) [Link]
It's mostly maintained, and maintained well, by Fermilab, who've just recently added a couple more people to it. It's also deployed on countless thousand of systems in the the LHC computing grid, as well as local physics department systems all over the place. It's future is secure precisely because of the state of academia - we need something just like it, we're not going to be paying for Red Hat Enterprise, and there's an awful lot of bright people involved; If FNAL stopped supporting SL, someone else it would pick it up again in very short order.
An awful lot of funding has goes into projects and experiments that currently rely on SL - without any serious alternative (no, CentOS isn't one), we can't and won't let it die.
CentOS 6 making its way out to mirrors Posted Aug 14, 2011 22:58 UTC (Sun) by iggymanz (guest, #78313) [Link]
You do know that Fermilab is largely shutting down in October of this year? There are some experiments and data analysis on site that will continue, but the main ones driven by the accelerator are EOL.
CentOS 6 making its way out to mirrors Posted Jul 11, 2011 12:21 UTC (Mon) by SEJeff (subscriber, #51588) [Link]
Perhaps you've never worked at a company that hired really good admins with the understanding it would be cheaper to hire 4-5 really good admins and pay them well instead of RHEL licensing?
CentOS 6 making its way out to mirrors Posted Jul 11, 2011 16:52 UTC (Mon) by robert_s (subscriber, #42402) [Link]
Will those 4-5 sysadmins generate new packages to provide you with an upgrade path when the latest RHEL derivative (WhiteBox/CentOS/ScientificLinux) stops providing prompt updates?
Just to clarify, I'm not saying "only use a paid product" - I'm saying I'd trust something like Debian (which has a whole community governance & infrastructure to ensure it does stay around - rather than having a situation where, for instance, some disenfranchised developer may decide to fork and make off with the domain name) any day over one of these distros.
CentOS 6 making its way out to mirrors Posted Jul 11, 2011 17:00 UTC (Mon) by SEJeff (subscriber, #51588) [Link]
Business solutions are not always technical. Somehow I feel you aren't grasping that. Perhaps the company has 0 public infrastructure? Perhaps they build an internal distribution based on CentOS or (crazy) Fedora for their hardened and locked down appliance? I also think that the people who blindly update to the latest RHEL/CentOS updates are somewhat insane. I for one would not trust my job on redhat not futzing an update that breaks things. Look at the RHEL4.6 kernel NFS updates. It was a real mess when they tried to backport the newer and faster google nfs client into their kernel. All of that work is now upstream.
I can tell you for fact (I used to be an admin there) that Ticketmaster.com has a metric shitload of CentOS servers. For their 4000+ servers, they don't have any issues with not using RHEL. I guess you just place emphasis on the distro where I place emphasis on the people managing it.
I don't trust smith and wesson to make my gun drop proof or safe all the time. I trust my fingers to not pull the trigger or drop the weapon when it is or possibly could be loaded. Given enough smart admins, most if not all security issues can be mitigated or stopped cold.
Feel free to rip me up. This is not trolling.
CentOS 6 making its way out to mirrors Posted Jul 12, 2011 16:33 UTC (Tue) by ThinkRob (subscriber, #64513) [Link]
I'm not going to flame you. That's a very fair, and well-reasoned stance.
There's something to be said for the "if it ain't broke..." school of thought.
Yes, I personally wouldn't trust CentOS for an Internet-facing box -- mainly because I don't have faith in their ability to reliably push fixes. But that doesn't mean that there is no valid reason to run CentOS. On the contrary, there are plenty of applications in which the stability, predictability, and binary compatibility matters, but the speed of security updates does not.
Ticketmaster is one such example of a place that needs many of CentOS's features, but apparently doesn't require rapidly-available updates. I can't imagine they're the only ones though.
The nice thing about the various distros is that each fills a niche, and they're not mutually exclusive. If you want stability and support guarantees, you can pick RHEL. If you want the stability and don't care about support or rapid updates, but need RHEL compatibility you get CentOS. If you want stability and rapid updates, but don't care about support and RHEL binary compatibility, you pick Debian. Etc, etc.
So yeah, even if they are late, CentOS serves a purpose -- just as Scientific Linux does, just as RedHat does, and just as dozens of other distros do.
|
|
||||||||||||