Surprise! Not ...
Is anyone actually surprised?
Just what did you think was going to happen after what happened with the CentOS 8?
Red Hat Enterprise Linux Server has maintained the same starting price over the past few years. Now, changes to the way the software is licensed have doubled the cost for some self-support customers using virtual machines. The change dates back to 2019, when Red Hat said its Red Hat Enterprise Linux Server, Self-support ( …
IBM have screwed up every software company takeover.
Lotus Notes -- was actually the best you could do and still have a workflow/db run on a 386. problem was IBM stuck with the original and hated UI for ever. Great idea destroyed by penny pinching.
Rational Software -- Rational Rose, Clearcase were ubiquitous in large companies. OO Superstars like Booch left almost immediately after the takeover. The products languished unloved for years.
Tivoli -- king of the entrprise monitoring software. Same old story of neglect, and, worse IBM's home grown collection of ancient tools (IEBCPY ffs!) re branded as Tivoli.
Redbrick -- the vanishing DB.
At my last place board level management required us to release on redhat enterprise.
So we had one licence left in its box. We did all development and testing on Mint, CentOS and RedHawk and told sales it would work on any Linux version given a particular minimum kernel and GCC/G++/clang version...
I never quite understood why people tie themselves to one variant of Linux distribution.
If you as a developer know what you are doing you should be able to do a single C/C++ build that works on multiple distros, given a minimum kernel, GCC and required dependencies installed (or shipped).
Best to look for better 3rd parties then...
Sometimes the choice of OS is merely an implementation detail. Company chooses a database, pays for licenses and support, and then gives the database whatever is necessary. This continues over the years until the cost of continuing is higher than the cost of switching to a different platform. This isn't just databases either: how often does an organization completely replace its phone system? Its monitoring system? Its accounting system?
"If you as a developer know what you are doing"
there in lies one of the problems. Developer think they know what they are doing, knowing what is required, understanding TCO and that someone has to manage/fix/maintain the system in 5 or 10 years time etc. Developers are lead by Project managers, neither of which have any motivation to think beyond the end of delivering the project, backed up by a company sponsor that wants the next shiney thing, none of whom undertsnads real world licencing issues.
Its a modern endemic development process, IBM are old school don't change anything unless it is obviously broken, evryone else is "Shiinneeey"
Personally I have never liked Red Hats Licencing model, it remoinds me of Oracle too much.
Oh and licencing is the only way to recoup the purchase cost, except when you are IBM.
We have IBM licence server to manage and report on 3 servers running Sterling Integrator, this is managed by a third party on behalf of IBM, who feel the need to have conference calls about once a month to discuss our licening position and probablly take a big chunk of the licence fee to cover their costs, to the extent that I doubt IBM make any money out of us (yes we are very over licenced so will likely cut it 50% when we renew so they willl lose money). Perhaps we should charge IBM for hosting thier licence server.
yep. not a fan of the end of centos, but red hats enterprise linux is what i know. so ive been sitting on a centos 8 vps slice waiting to see whats happens. im not interested in stream, oracle is too evil even if their rhel clone is free but today i used the alma linux beta script to convert it (alma linux is the free cloud linux version) and it worked perfectly. im sure other rebuilds will be good, but alma is good to use now.
Then don't use openldap. It's awful, especially as packaged for RHEL. Even though creaky old, RH's own 389 Directory (derived from the classic iPlanet/Netscape DS) is much less challenging to set up and maintain over time. But also work towards eliminating the LDAP protocol from your environment. Pick a web API (or several) and run with it. Backend with a reliable and uncomplicated document db. As for RHEL itself, just because IBM has set itself on fire in a fit of greed doesn't mean your company has to. Take a serious look at Ubuntu and SuSE for solid, commercially supported, distros outside the cloud. In the cloud you've got even more choices, like AMI. If that's incompatible with your legacy Oracle and other enterprise shelfware, maybe its time to look at ditching them. The self-immolation of RHEL comes at a pretty opportune moment, actually, as workloads move inexorably to containers and container orchestration where what distro is actually under the hood matters less than ever.
So you're moving from an enterprise Linux distro with a decade of vendor support and it's free clone to a community Linux built from stale packages and a proven track record of fuck-ups?
I certainly understand why people leave the RH platform after the legacy CentOS termination, however most people using CentOS did so because of the stability. Which comes from the fact that CentOS was a 1:1 clone of RHEL, which is an enterprise grade Linux distro which is maintained mostly by professionals and which has quality assuring measures in place to ensure the OS is, actually, good enough for enterprise class computing.
Just looking at Debian's quality issues over the years (including the massive openSSL fuckup) gives a horrible picture. And the fact that the "LTS" (5 instead of the standard 2 years) "support" comes from different and much smaller community doesn't exactly instill confidence.
If stability and security is a concern then it makes a lot more sense to either set on one of the new RHEL clones like RockyLinux, or move to a different enterprise Linux platform like SUSE/openSUSE.
You're looking to move to something with a longer track record than Red Hat's entire existence. To a distribution that is also upstream to several projects, maintained on multiple architectures and with a faster MTTR on security. Oh, and also the basis for 200+ other distributions. ...
Red Hat - five years [ of vendor support] for older packages, maybe, and no free clone. As Red Hat moves away from being a Linux provider - why _not_ go for a community supported distribution with the same underlying systemd as Red Hat - and a choice of multiple desktops?
"You're looking to move to something with a longer track record than Red Hat's entire existence."
How about MsDOS then?
Besides, a long track record doesn't mean it's also a good track record. Debian is a good example of that, it may exist longer than RHEL/CentOS but the Debian community has shot itself in both feet so often it's a bit of a joke, really.
"To a distribution that is also upstream to several projects, maintained on multiple architectures and with a faster MTTR on security. Oh, and also the basis for 200+ other distributions. ..."
Besides that Debian's MTTR in regard to security is pretty abysmal compared to enterprise vendors like RH and SUSE, what exactly do you think being "upstream to several projects" brings to the table when the main aim is stability, reliability and security?
Or that it is the basis of 200+ other distros (of which most are aimed at home users and of which almost all share the exact same problems as their baseline distro)?
Debian is fine if you enjoy fiddling in the innards of a Linux distro or if you want to roll your own stack, but if that's what you want then RHEL/CentOS has been the wrong choice even before RH killed legacy CentOS. Same as Debian is the completely wrong choice as a replacement for a proper enterprise-grade Linux distro.
You have clearly NEVER used Debian professionally, right?
Some data points:
- I still have nightmares about "RPMhell" from years ago. Last time I tried, the situation with dependencies was better (since RH cloned Debian's APT in the form of YUM) but tended to explode at the worst moment with no way back.
- The unofficial slogan for Debian happens to be "...when you've got better things to do than fix systems"
- Huge ISP/MSPs depend on Debian... that's not a coincidence.
- Most full- or half- time Debian Developers are paid by companies to do so. And specifically not that many are paid by Canonical, by the way...
- RedHat's support really isn't good. Enough to CYA, yes. But it doesn't actually solve very complex problems --- i.e. same as Cisco's TAC, only a bit better
there's a reason SuSE makes *loads* of money selling support for RHEL (and they really aren't much cheaper)....
- QA for packaging / overall distro stability and Debian's own software IS actually top notch:You''d believe that when ~2000 people spend about 18 months stabilising and polishing the overall distribution and how it fits together, the result should be at least "decent"... and that's not considering the over ~20k packages in Debian vs about ~3K in RHEL (or Ubuntu).
And well, RH's insistence to force their sometimes ill-conceived software ---I personally *like* systemd--- like SSSD and all kind of dbus concoctions down your throat, only to "differentiate" (read: make you actually need their support) from the competition.... well, I'd rather not.
But YMMV, of course
The kernel and significant parts of userspace have a licence which explicitly states that placing additional restrictions downstream is forbidden. Such as that 'no commercial use' thing for the free of charge version.
Of course Redhat are a significant presence in the organisations that try to get compliance...
Qt also has a commercial model which trys and does confuse people.
Only a few of the Qt libraries are GPL and can be completely avoided.
The sales people say if you develop using Qt under LGPL you have to pay them a large sum of money before you start using the commercial version. There is nothing in the LGPL that supports this...
Given that we only use it for the GUI it's very expensive when compared to a full msdn developer licence.
> The kernel and significant parts of userspace have a licence which explicitly states that placing additional restrictions downstream is forbidden. Such as that 'no commercial use' thing for the free of charge version.
Maybe so, but if you're planning to rely on that, what you're actually doing is saying that you accept the (very real) risk that you may end up having to face-off against IBM's legal team (known as the Nazgul for good reason) and the business will bear the costs of the resulting prolonged and expensive litigation..
Whether or not you're right in your interpretation, no sane business should take that position. Better and safer to move to another distro and simply mark RHEL as a no-go.
Red Hat specifically do *not* sell you the GPL protected kernel etc. What they *do* sell you is support.
If you use Red Hat extensively, as we do, this is made very clear.
In fact, it's even baked into the terminology as they are selling you Subscriptions or Entitlements, not Licences.
All the licences apply to the various bits, I would imagine some of the more recent distributions could have many libraries and widgets and bits bob that are covered by a whole raft of different types of Open Source licences, that would be as long and most Oracle EULA statements. Let's be honest here, the Open source community has not really done itself any favours with all these different licences for the same things.
This is one of the problems with Linux, the Kernal is good, the bits that get added on are of variable quality, have their own support update issues and is a bit of a mess really and use a mass of different licence models. It's no wonder teh likes of big business is looking for a one stop shop, or if you are really big, run your own custom distro. Imagine Facebook or Amazon suddenly finding it need to pay double its support bil?
If you sit at home on you Linux box that is not really an issue, if your $1bn business relies on it and you may get stung for massive licence/support costs or someone may take over your favourite small fry distro and try and monetize it, well the relative peace, stability and certanty of MSland does start to look apeaaling
i dont think thats true, they just need to supply the source code including their changes and improvements, which they do, and they have to accept the existance of clones, which they do. if they want to push their users further away, so be it. my recommendation? c8 -> alma linux.
Into the loving arms of Debian/Ubuntu
No, NOT Suse - unless you like being thrown under a bus when things get difficult
Suse have form for "abandoning customers with difficulties") and are in MY personal "No business dealings EVER until the heat death of the universe" list. I'd sooner deal with Microsoft than Suse.
Redhat Satellite seems deliberately _intended_ to get rid of enterprise customers by making it too difficult to get anything done(stuff which used to take 10 minutes is now a week-long ballache instead)
May you please tell me more about your comment on SuSE's support?, I'm really interested to know.
Where are you located?, when was the issue? was your bad experience with SuSE support or with one of their partners?, what was the problem?, what happened?
Sorry for so much questions, i live/work in Bolivia and whenever opened a case, they answered from their HQ in Argentina, slower than their competition but answered and helped to solve the issue.
I want to use more SuSE but willing to hear others experiences before advancing.
Biting the hand that feeds IT © 1998–2021