[olug] Centos

[Christopher Cashell]

On Wed, May 22, 2013 at 9:37 AM, Sheldon, Roger W
<Roger.Sheldon at firstdata.com> wrote:
> Company I work for is currently using SLES but looks like new builds will be Centos for non-critical. Looking at their website support/ticketing perspective I don't get a warm fuzzy.  I believe it's RH w/out the $$ cost or support. I think Openlogic is the 3rd party vendor support we are going with. Would appreciate input from anyone using Centos good bad or indifferent.

A lot of this has already been discussed, but I wanted to address your
original questions a little.

First, your assessment is pretty accurate.  CentOS is a
binary-compatible repackaging of Red Hat Enterprise Linux (RHEL).  It
looks and acts essentially identical (any functional differences are
considered a bug), just without the name and trademarked branding.
The CentOS gang do not offer any direct support, although there are
third-party vendors that offer support for it.  Whether there are cost
advantages of third-party support vendors over Red Hat (using RHEL), I
don't know.

For business use, I'm a fan of CentOS.  Red Hat does a good job of
targeting business and enterprise use with RHEL, so there is a lot of
functionality to support that.  Third-party software almost always
targets RHEL (and possibly others), making it a good platform for
running external software.  There were some issues with CentOS updates
lagging (significantly) behind RHEL a few years ago, but the CentOS
guys have done a good job of keeping things up-to-date recently.

At a previous job, we heavily mixed RHEL and CentOS.  Basically, our
most critical and key points were run on RHEL, while the majority of
the infrastructure was run on CentOS.  We considered CentOS to be our
base platform, and RHEL to be the platform for third-party software
requiring a supported stack (such as Oracle), and for servers that
couldn't easily be replaced (ones tied to expensive SAN storage).  We
did use CentOS for critical production use, but primarily in cases
where an issue wouldn't cripple us; for example, in a large back-end
processing farm where we could easily swap in and out new processing
nodes.

> Roger Sheldon

--
Christopher