Back to Blog
Wakeonlan centos7/10/2023 ![]() For anyone doing large scale virtual hosting on CentOS - good luck finding people to pay for the increased licensing and real costs of support for their VMs. For universities / HPC clusters / ISPs and hosting providers who've been relying on CentOS because Red Hat licensing is, frankly, painful for large numbers of machines (and almost no-one really understands how to provision Red Hat correctly using Satellite or Spacewalk) - they're the community that will lose by CentOS losing long term support (but are also the sorts of folk to have sysadmin staff doing their own custom support). For Red Hat to support a large customer - a bank, a single government ministry, the US DoD - they effectively have to do discovery exercises, build specialist support teams and hope fervently that they don't find too many "specials" crawling out of the woodwork. For everything else,you have to rely on EPEL or other third party repositories (at which point potentially you lose Red Hat support)Īt that point, just like with Oracle database support, you also find that "Oh, your xyz system is not quite our xyz now and we can't support the customisations your local admins have made" and you're on completely on your own again. In my (limited) experience with the Red Hat ecosystem, senior management and accountants rely on the apparent promise of support you get for a distribution over ten years - and large projects take five years ++ - but, in practice, developers and power users need something new after about two years when $NEXT_BIG_THING doesn't work on Red Hat/CentOS By comparison with Debian/Fedora/Ubuntu, the Red Hat/CentOS subset of Red Hat supported software is tiny - it's ideal for your long running services but doesn't support much software natively. In another one of the threads on CentOS, Maniax said - below a certain size, you use a distribution, above a certain size you build your own distribution. I think the illusion of support is the thing that resonates with me. TL DR: due to some industries' established practices, obsolete LTS distros set in stone (and no-license-fee ones, at that - leeching is normal behaviour, and the recent change for CentOS has put some people in panic mode) remain very relevant :'( * the security support is limited to a fairly narrow set of packages and isn't necessarily easy or meaningful in the long term: few upstream projects perform security backports to versions more than 2-3 years old, let alone 5 years old. However, these people might be forgivable for not being aware - after all, they're using CentOS 6 (what EOL in a few months, exactly ?), if not earlier ![]() * these distros have already started changing with the times, as shown by the growing number of backports from minor release to the next one, in the late CentOS 7.x series and the CentOS 8.x series. In some industries, buying and deployment cycles are extremely long (5+ years) and costly, and many people remain under the illusion of stability and security maintenance provided by LTS distros.
0 Comments
Read More
Leave a Reply. |