On 07/23/2011 11:35 AM, 夜神 岩男 wrote:
On 07/24/2011 02:14 AM, Yasha Karant wrote:
A vendor professional systems person whom I know has been requested to
install SL 6 on a system that is being configured for us. In a
discussion with him, he gave me the opinion that his (vendor's)
experience with SL is that it is "buggier" than CentOS, and CentOS often
"fixes" RHEL bugs.
They both fix bugs when found and both communities are fairly good at
pushing those fixes upstream. I find the documentation of bugs in SL
more concise and helpful than in CentOS (seems the ones submitting bugs
are less prone to freaking out at their computers or submitting SWAGs to
bug trackers). There is definitely less hand-holding within the
community -- and a lot fewer requests for hand-holding from what I've
observed.
For anyone on
this list who is familiar with the post-RH release handling and
qualification/testing procedures of RHEL source by either or both
organizations, or by the Princeton University distribution of RHEL,
direct comments would be appreciated. Is there any factual data,
including procedural differences, to support the opinion that I have
been given?
I am not aware of any actual test data that compares the various RHEL
derived distros under any stress in a meaningful way (are you
volunteering?). I have deployed RHEL6, SL6 and just recently toyed with
CentOS6 test deployments, and found not enough difference to warrant
including CentOS in my thinking for now (for non-technical types with
deadlines RHEL is worth the money, though).
Anyway, CentOS 6 was just released the other day -- it hasn't been out
long enough to compare or for deep, weird problems to be uncovered yet;
SL6 is fairly well understood at this point. But, name recognition alone
goes a long way to framing most people's interpretation of software (and
other things), so bear that in mind when listening to people.
From a non-technical perspective, however, there is a huge difference.
The SL6 community has far fewer knotheads than the CentOS community, and
accordingly less drama. It also feels easier to find things, though I'm
not quite sure why (fewer meaningless articles laying about?). In fact,
I've never had to ask a question on list before. SL is therefore
considerably less buggy as a community. The frustration index is a lot
lower with SL6 in other than social ways. The development and release
process is a lot less mysterious than CentOS, for example, and this
makes planning a little easier.
tl;dr: No hard data to support your friend's claim. SL6 is lower on the
stress & drama scale.
-Iwao
I certainly agree with your sentiments concerning support via the CentOS
community, and even more so from some of the other Linux distributions
that are not aimed for stable production but rather for either
enthusiasts or casual users. The issue is that many of the CentOS (not
to mention Ubuntu ... ) users are typical end-users, used to the MS
Windows or MacOS environments, and some from business vendor supported
systems for which source is not available and with (more or less) full
paid technical support -- typically a management information systems
background at best, not a rigorous computer science or engineering
background (MIS: how to use an information system for the mission of an
organization, typically financial gain; CS: how to develop and build
the fundamentals upon which information systems can be deployed).
To be clear: this is not a "friend", but a high ranking engineering
person at a vendor who was presenting his personal, or his firm's
internal, expert opinion. My posting was an attempt to discover any
data relevant to the opinion, as well as to at least get the
distribution release testing policy/methodology for SL (and perhaps
CentOS if there were any CentOS organization persons on the SL list
willing to comment) once there is a release of source from RH (source
that RH is required to release).
As to your query: I am not aware of any actual test data that compares
the various RHEL derived distros under any stress in a meaningful way
(are you volunteering?) -- regrettably, I lack the staff (because of
funding) to undertake any such systematic comparisons, or even detailed
stress testing of whatever distribution currently is being deployed (on
this laptop, CentOS 5.6, whereas on my workstation and on our new
research cluster GPU compute engine, SL 6.x, x being kept current
production). Thus, my experience is anecdotal, and in terms of the
actual process of deployment of either SL or CentOS (or RHEL for that
matter), an outsider looking in and not seeing much past the "magician's
robe" until the "rabbit is pulled out of the hat" -- that is, a distro
is released in any form (beta or production). I was hoping that at
least for SL, the curtain would be lifted as to how RHEL source is
converted into a SL production distribution (if this is documented, the
relevant URLs would be appreciated).
Yasha Karant
