Scientific Linux 6.0 released [LWN.net]

Scientific Linux 6.0 released

Posted Mar 4, 2011 16:28 UTC (Fri) by southey (guest, #9466)
In reply to: Scientific Linux 6.0 released by Np237
Parent article: Scientific Linux 6.0 released

You are welcome to have your definition of scientific usage but don't force others to have the same definition!

Yet many people like myself are successfully using Red Hat, Fedora and related distros on scientific workstations. Personally I prefer missing components because distros usually don't have the latest version (Debian experimental is not an option) or that I want to know what I have installed rather than a huge bunch of unused dependencies just waiting to cause havoc. Also some scientific packages (blas/lapack) have a history of being broken or incomplete so it is hard to automatically trust some distros.

Scientific Linux 6.0 released

Posted Mar 4, 2011 17:11 UTC (Fri) by donbarry (guest, #10485) [Link] (5 responses)

Odd to hear how the *lack* of components is spun as a benefit.

I am an astrophysicist, and also the computing architect for
a NASA spacecraft instrument team. I migrated a team from
Sun workstations to Redhat 7.3 eleven years ago, and stayed
on the Redhat treadmill to RH9, while slowly learning the many
good things about the Debian system. With the mischief and
misdirection that followed RH9, I professionally migrated to Debian.
I've been there ever since.

Debian has a release cadence that is reasonable for scientific
teams, and it packages *far* more software than Redhat. Although
both systems make it easy to maintain customized repositories
of super-versioned software, the fact that much of it is already
in Debian means that one doesn't need to reinvent most of the
packaging decisions.

Furthermore, one can trust, with full integrity, that the underlying
distribution will not jerk you around in pursuit of enhancing their
revenue stream. Redhat has done that several times in the last dozen
years. With a new management team and an overvalued stock, they will
do it again in the next few years (see the recent kernel patch
distribution discussion). You have been warned.

Debian's decision system comes from -- the community. A *real* community, not just one assembled by corporate fiat, or temporarily bribed into existence (a la Shuttleworth). I trust that Debian's integrity, the heft of its package choice, and its direction will remain constant for another dozen years.

The "Scientific Linux" system is an impressive effort from Fermilab, but the decisions as to its architecture base, inevitably, were made by bureaucrats, not scientists. That's simply the way the large national laboratories work, when you seek internal funding to support multiple personnel for a multi-year effort. Of course, scientific guidance would have been sought, and given, but inevitably compromises between scientific interests and bureaucratic interests emerge. It is not at all surprising that RHEL was chosen as a base, given the natural tensions. A scientist
would advise: "RHEL is acceptable as a platform, if possibly not ideal, given the limited packaging of scientific software. Debian is better."
A bureaucrat would rationalize: "I can defend the RHEL choice based on
the substantial and growing revenues of Redhat and its contractual
adoption by major industrial figures. If SL fails and we had chosen
Debian, others may characterize it as putting the prestige of a
national laboratory atop a volunteer, hobbyist effort, and I will lose
face. My interests are served by the commercial effort, which
partially absolves me of risk."

Scientific Linux 6.0 released

Posted Mar 4, 2011 18:29 UTC (Fri) by kmccarty (subscriber, #12085) [Link] (2 responses)

It was actually suggested at one point (by a Debian user) on the SL devel mailing list that Scientific Linux rebase themselves on Debian:

http://listserv.fnal.gov/scripts/wa.exe?A2=ind0703&L=...

Not too surprisingly, there weren't any replies.

Scientific Linux 6.0 released

Posted Mar 4, 2011 19:08 UTC (Fri) by donbarry (guest, #10485) [Link] (1 responses)

I suspect that comes from two factors:

1) Not many scientists outside of subdisciplines of physics use
"Scientific Linux". I do not know a single astrophysicist who
does. I'm sure there are many users at a few particle physics
labs, probably dictated by management.

2) Those who know the inertia of bureaucracies figure they have better
fish to fry, and as long as someone else is packaging the software
they need, would prefer to tilt their lances elsewhere.

That said, I think that the "Scientific Linux" packagers would be
very well served by rebasing their efforts onto Debian.

Scientific Linux 6.0 released

Posted Mar 7, 2011 17:36 UTC (Mon) by kmccarty (subscriber, #12085) [Link]

Coincidentally, mention of Scientific Linux just came up on the debian-science mailing list. Some highlights:

http://lists.debian.org/debian-science/2011/03/msg00006.html
http://lists.debian.org/debian-science/2011/03/msg00007.html
http://lists.debian.org/debian-science/2011/03/msg00012.html

- Kevin

Scientific Linux 6.0 released

Posted Mar 5, 2011 19:19 UTC (Sat) by Np237 (guest, #69585) [Link]

We went precisely through this in my company. It was very hard to get Debian accepted while the company already had Red Hat as the primary OS for Unix servers. The first attempts were made with Red Hat, with mixed results. Those factors made Debian win:
- A number of motivated people explaining over the benefits to management over the years.
- Money money money; we can do with a team of 10 what a team of 20 would not be able to do with another OS.
- The universal operating system: we can use the *exact* same OS, only with differences in installed packages, for scientific workstations, computing servers, and HPC clusters. Binary compatibility between them saves a lot of money too.

Scientific Linux 6.0 released

Posted Mar 10, 2011 5:41 UTC (Thu) by dowdle (subscriber, #659) [Link]

Let's see... you haven't used RHEL at all... because you quit using RHL about 8 years ago and you are still complaining about what happened back then. You are positive that Red Hat sucks and will continue to suck... and offer a warning... that it will never be good enough because it has commercial interests... although it just happens to be the shining beacon of free software and the leading example of a financially successful commercial Linux provider... who has done a rather good job at being serious about and sticking to its FOSS principles... buying up several commercial companies/products and releasing them as open source... and being a significant contributor for years to the kernel, gnome, gcc, xorg, etc. As far as I can see, it is impossible for any commercial Linux to be worthy in your eyes... no matter what.

Notice how I did not denigrate Debian in any way as I do not feel the need to. I have a lot of respect for its developers and its users... and I certainly appreciate the competition it provides with its good work.

I have participated in many "distro wars" type conversations over the years but I'm beginning to grow weary of them. There are many use cases for Linux and no single distro that meets everyones needs.

Red Hat doesn't have a smaller package count than Debian by accident. As you may or may not recall, they intentionally reduced their package count going from RHL to RHEL in order to make it easier to support in a commercial environment. As you are surely aware, they support their releases for a long time... 7 years at least. If I remember correctly, Debian supports their releases for 1 year after the next release is made available. That is a significant difference that may or may not be important depending on your needs.

Please refrain from telling everyone that if they pick Red Hat over Debian that they are doing so because they are forced to by bureaucratic managers. While that might be the case in some instances, it certainly isn't in all.

Also be advised that your suggestion that Debian has the only true community... is offensive to me... and probably others. Fedora has a community that is sponsored by Red Hat. They do a lot of work. Like Debian's community, Fedora's is often consumed with lively debate and disagreements... an actual sign of a healthy and honest community. Red Hat has a community too... and they include users like myself... the CentOS community... and the Scientific Linux community... among others.

While I appreciate your commentary and the thought you put into your comment(s), I think you should be a little more open minded. Debian is a great distribution and great bunch of packagers, but you yourself benefit by a lot of the development contributions done by Red Hat even if you don't choose to use Red Hat's packaging of them.