Hi,
I’m sure you all heard the news and the entire Linux world in under shock.
RedHat pulled the plug on CentOS …
Now this is an Opportunity for SLES ! But SuSE must listen and think about this Huge opportunity.
The founder of CentOS was quick to react and is already making a very very Clever move:
Starting the Exact same project but with other name: https://rockylinux.org/
I’m certain Rocky Linux will take the majority of installs away from “CentOS Stream” and will be made a default OS on most dedicated servers and Hosting services.
I’ve seen mentioned on several openSuSE meetings the possibility that I think it is now a Huge Market opportunity to make available the same CentOS project but for SLES.
If SuSE is smart and knows how to remain relevant and even increase their OS market share …this is It!! Time to allow a CentOS version of SLES with the exact same long term support …
People moving away from CentOS alone would give an instant Boost to SLES-clone market.
Time for action is now.
My dear friend I beg to differ … I will always called it openSuSE. Been using old SuSE versions since 1998 … so old folks like me tend to stand by their principles …
but of course anyone calling openSuSE openSUSE is absolutelly free to keep their forged designation …
Now, what do you think about the issue … the opportunity?
So using SuSE for about three years and then SUSE (and maybe even openSUSE)) for another sixteen never learned you to adapt? We can be glad you never tried something from S.u.S.E.
What do you expect to do from me and the other openSUSE users that populate these openSUSE forums?
If you can get yourself over your disgust in having to go to a company of that name.
I’m sorry you could not get over my nostalgia … I truly did not mean to upset you or anyone involved in the distribution I consider the Best for many years.
But can we not discuss in here this subject?
I would be interested to know what is your position on the fact that CentOS is going to pulled out of their current status as the RH stable clone into something quite different.
And also what this could mean as a huge opportunity to SUSE distro(s)?
Thanks.
While I feel pity for those who lost the distro they used (I only have to think about my own disappointment when openSUSE would be stopped, bringing all the scores of having to find another and getting used to it), I have no further position. I wish all those hit by this wisdom in their choice for the future.
Redhat was bought by IBM - they want the revenue so they are killing CentOS. I saw that coming a while back. I was surprised that it took this long for it to happen.
I hedge my bets with both Ubuntu and openSUSE. Each has pluses and minuses. Ubuntu is way ahead in ARM support. I like openSUSE for x86_64 but Ubuntu has many more packages for it than openSUSE.
It is sad that the linux community has so many installable file formats that are not universal. those that claim to run on any platform are so large that your computer runs so slow.
AT&T unixware had packageadd and packagerm to install or uninstall software. Everyone else had to improve that - rpm was pretty good to replace it. then there is deb.
I guess it is like Democrats and Republicans for which format is better. Too bad we cannot do what is best for all even if it is painful.
I used init from 1973 - now I accept systemd as the replacement - it was painful as I had a bunch of init scripts.
Hi
How so, you mean the usual Ub* hacks rather than following upstream (or submitting upstream)? I find that interesting considering ARM employees are involved in the openSUSE ARM arch…
I found the transition pretty simple, not without some pain, one thing is timers…love them begone cron!
I have some Raspberry Pi 4 8gb.
Everything works with the Loboris hack of Ubuntu. Sound, video’s, ethernet, wifi and all the apps I use in OpenSUSE. and so does Pi-hole. I could not get many things working with OpenSUSE and the MATE desktop - I dislike Gnome3 and KDE with a passion - If I wanted Windows - I would use Windows - I hate their new interface. Long Live Gnome 2 as MATE.
Hi
Ever thought of heading to IRC #opensuse-arm or the openSUSE ARM mailing list to ask? I’ve found IRC the best medium, some of my earlier issues with RPI3’s where resolved in no time at all.
AFAICS, Red Hat talks about “a long period of support” …
CentOS 3 – released in 2004 – full updates until 2006 – maintenance updates until 2010 …
CentOS 4 – released in 2005 – full updates until 2009 – maintenance updates until 2012 …
CentOS 5 – released in 2007 – full updates until 2014 – maintenance updates until 2017 …
CentOS 6 – released in 2011 – full updates until 2017 – maintenance updates until 2020 …
CentOS 7 – released in 2014 – full updates until 2020 – maintenance updates until 2024 …
CentOS 8 – released in 2019 – full updates until 2021 – maintenance updates until 2021 …
IOW, “about 10 years” – maybe …
Lets face it, that’s an awfully long period of time – these days …
In the telephony world, it used to be that, (mechanical) switching equipment was expected to have a life of 30 years and, the financing was set up on that assumption …
In the mobile telephony world, the cost of a Base Station – the box attached to the sector antennas at the site with some heavy real-time computing hardware in it – is often recovered within 3 months of service …
[HR][/HR]Therefore, the question needs to be asked, the way things are at present, is it realistic that a CPU and the related hardware will be in service for 10 years?
[HR][/HR]With the possible exception of Avionics and Auto-Pilots – especially military aircraft but often, also civil aircraft …
Air surveillance systems – Radar – are also in this category …
Yes, in these areas, the people in those industries just love the “cosy warm feeling” of “being able to live with a single Operating System release” for the life of the maintenance contract – it makes life “just so easy” …
For the rest of the world, we tend to buy new hardware on a cycle of much less than 10 years – mobile telephones are an obvious example – and very rarely attempt to run the previous operating system on the new hardware – which the old version of the operating system usually doesn’t support anyway …
Yes, we all migrate our data from the old system to the new system but, that’s another matter and, it usually doesn’t raise too many issues …
My main issue with every RPi alternative to Raspian are their reliance on Uboot…
And that uboot is so enormous and internally complex, it’s impossible to do much beyond the simple command options available on boot.
So, for instance Raspian has this really nice, easily extensible support for hats.
Try to do the same thing or port to anything running uBoot is impossible, so all that extensible hardware is inaccessible unless you run Raspian.
I believe that uBoot badly needs a re-factoring to simplify its code and build new APIs, particularly those that support “makes sense” features in Raspian. But, it’d be an enormous project, probably too big without substantial sponsorship.
Reason why Raspian boot can’t be used elsewhere is that it contains proprietary code, is not publicly licensed.
I’ll be watching with interest if Rocky Linux gets off the ground…
For many (6+ ?) years now, from afar I’ve observed CentOS make less sense all the time as technology has rapidly accelerated logrhythmatically.
A prime example is what the Linux kernel has become… Compare the number of changes and the release schedule of features 10 years ago to today. Used to be, we might see a handful of “important” features each year, maybe two. Today, a new significant release is every 2 months where even a minor version regularly introduces major new developments, every one packed with “necessary” support for new hardware, software, patches, and so on.
For a distro like CentOS that literally hung its hat on the stability of using the same kernel up to 5 years, how is that possible anymore even with a massive effort at backporting? Even if security patches were backported, how does the distro stay relevant without supporting all the new hardware that appears all the time with Linux the choice for more and more mobile devices and system boards all the time? It used to be that small factor system boards were launched every year or two, today everyone who has a new idea combination of features is launching a system board within a few months after design.
Looking further into the future, I can see how even a “conservative” release schedule we see today of LEAP, RHEL, Debian and others might become outmoded and rolling releases might be the only model that makes sense.
So, sorry to all the people out there who still want an LTS today…
LTS might not make sense even today, and will likely make even less sense in the years to come.
My guess is that LTS might be replaced by better tools that improve the reliability of rolling releases like what openSUSE has been doing by being at the forefront of implementing technologies like BTRFS with snapshots enabled by default (Even Fedora doesn’t seem to implement snapshots by default AFAIK, and Arch doesn’t implement snapshot recovery by default using Timeshift) and Transaction Server. In other words, I think that SUSE/openSUSE has got it right by understanding the trajectory of technologies and why some are essential over others… Something I don’t see obvious in other distros.
That’s were I truly do not agree with your argument. It does make all the sense to support an OS for at least 10 years if not more. Specially for companies with hundreds of systems all doing Exactly the same Thing for decades: aka Servers. Hence the name Server OS.
But you bring two types of issues:
Hardware lifetime vs OS lifetime.
The all point of using Linux is that They should be Completely Unrelated to the end user.
-> That’s the Key Selling Point on Linux. Runs Everywhere Forever!
On 99% of the worlds Super-Computers with 5 Million compute nodes+ to 90% of smartphones, the Mars Rover, 90% of home routers, 99.9% of all TV’s, Nuclear Submarine … ya name it .
But in particular for the Server Market: the Value for the customers that pays, specially the ones that pay a lot to have Linux on their Servers is Exactly that … Long Lifetime Support! That’s Absolutely Key for them and for their decision. That’s why Linux companies like RH and SUSE have Such a huge value proposition on the long run.
And OS Support that Should live Multiple Hardware platforms is Critical. That’s a Huge Cost savings. And Above all it’s not only the cost savings of Software alone it’s the entire Operational Environment of those businesses.
Imagine a company say like google saying …OK … we have to change all our 1Million servers cause … RH decided CentOS 8 it’s over in 2021 … (I know I know they have their own customized and house maintained version … but it’s just an extreme case scenario).
Try to run something on a windows server OS from 10 years ago … check how many platforms they have discontinued …how many customers Changed from windows to Linux Exactly because of that.
And Hardware life expectancy is truly not and issue for those customers at all.
Also I don’t agree the lifetime of a server is typically 5 years (the industry very hill-explained average) simply because we are now entering a time when incremental gains on Hardware speed/performance will be diminishing the example is this wave of new AMD CPU’s that are fabulous … but the truth is the gain they bring is not even close to former generational jumps of the past.
ARM can have another leap forward but mainly on the energy saving side of things … increasing even more the Core count for the same energy consumption, maybe there we will find a bigger leap forward … but I don’t see a lot of investment other then evil apple on that area.
The typical 5 year old server is seen on Generic datacenters … those that make colocation for web-hosting and root servers, VM’s and the like.
For those hosting companies yup, they must flip the server every 5 years since for them saving a bit on having more compute density or maybe less energy consumption is very important.
Otherwise a server hardware is expected to live for 10 years if not more on most Businesses … And Software much longer then that …
I love to update my laptops and desktops with multiple versions of all kinds of Linux … but that’s my computer … my “tinkering” machine. And yes every time I get the chance I try to upgrade the hardware.
But that’s not the case for servers.
Also note that base station telecom hardware is seating outside under all sorts of bad weather conditions … so when someone from the telco’s tells you a Very Expensive base station on a Cell tower is changing every 3 Months …something very wrong is happening with that base station …
That’s truly Not the Case for telco equipment.
To finish my arguments I have to remember everyone How Linux took over the world of web servers … yes … CentOS played a Fundamental Role on that — that was one of the Key Obvious Processes of Datacenter Domination:
Free Long term use Linux Server OS.
Just check the statistics of what runs on the Web and it’s clear … Linux Dominates the Web -Server Market
Some markets are almost 100% Linux-Based … DNS Bind, Web Servers … you name it.
And that was made possible largely due to yes … de distribution that was used as a OS-Commodity: CentOS
Time for SUSE to wake up and understand the Basics of DataCenter Domination …
The main issue mobile Telecommunications faces is, the rate at which the (mobile) Standards change – in 2020 the 3GPP meetings slowed down but, there used to be quarterly meetings to approve proposed changes …
The major international standards – 2G, 3G, 4G have all had major changes with respect to data throughput over periods of only a few years – much less than 10 years …
For other country specific standards such as CDMA and W-CDMA, it’s much the same …
I expect that, 5G will also be “quite unstable” – especially during the initial rollout …
It is not unusual that, during the lifetime of any mobile telephony standard, hardware changes are needed for all the components which make up the network.
The fact that, typically, the financing of Base Station sites is often covered by the first 3 months of operation, simply means that, the companies operating the networks can easily afford to upgrade the hardware as needed to meet the data throughput demands of their (mobile) customers …
Their customers, seem to be able to frequently purchase new mobile telephones to take advantage of the increase in data throughput being offered by the mobile network providers – it’s a continual revenue earning machine … >:)
I agree the rapid change tech in those equipments is a reality, But most countries are not yet covered with 4G … some not even 3G.
And Above all:
2G started 20 years ago, almost all of the base stations I connect to have full 2G support. Trust me they will not replace the hardware unless there’s a failure.
3G started 14 years ago or more and All the base stations …all the base stations I connect to have full support for all those 14 years+ Actually when there’s a drop on coverage on 4G or a failure on 4G the default is 3G …
4G … same will happen …
5G is going to take years, Decades! It requires Way more antennas to operate to implement and Still all those previous Tech’s will be Supported … 3G, 4G is not going away anytime soon. And yes people expect to have full 4G and 3G support in the next 20 years.
Changes on the base equipment does not meant changing all 100.000+ base stations on a entire country …they are all Expected to last at Least 20 years. New tech means Added hardware … Almost Never replacements.
Base Station Keep Growing with Added Hardware … and although of course there are needed Hardware Stations Hardware upgrades the Clear specification for Any Telecom Equipment by Law is that it must Last and pass All compliance tests to make sure the hardware survive at Least 20 years … with Full Software Support … even if it is replaced before of that.
Replacing Hardware on a base station every 3 Months is not normal at all, adding Hardware every 2-3 years can be … replacing hardware that frequentelly it’s for sure a Huge problem with damaged enclosures … not equipment replacement per se.
Oh yes and the market for more towers and Much faster bandwidth is in huge demand. Many decades ago everyone with knowledge in technology could foresee the need for such an increase in demand. There’s Nothing like bandwidth available on your pocket … that I also know … but the rate of change is actually the main reason Most customers complain … atrociously slow.
Linux does run on practically all hardware… nearly forever but things have been gradually changing over recent years.
Since practically all hardware support is now distributed through the kernel, practically the most onerous problems related to setup on common hardware have been eliminated… Just detect the hardware properly and the driver is immediately available, installed and configured.
But, with the continuing explosion of different types of hardware, maintaining support for every hardware that has ever existed is quite an undertaking, so from time to time support for some devices like popular graphics cards in the 16-bit era have been deprecated. When this happens, you might need to explore qemu emulation solutions.
The more common problem though is that today’s global threats are different than what existed pre-Internet and absolutely nothing including Linux can run for very long without active security management including patching. It used to be that some Admins were proud they hadn’t touched their servers in years… If someone would say that today, it’s almost certain that machine could be owned as soon as people knew it was unmanaged.
So, the whole idea of running machines with little attention is making less sense all the time, and planned upgrades have to become part of your security strategy… Both systems and subsystems have become nearly impossible to deploy and maintain properly without upgrading to whatever is current and stable which makes the LTS philosophy harder to implement all the time (and therefor more expensive in effort and time).
openSUSE was probably at the forefront among distros re-organizing to decrease effort (which translates to cost savings, reliability, timeliness and maybe a better product) and it looks like although late to change, RHEL is adopting this same model long after Debian which came after SUSE(both affecting downstream and sister distros).
So, the User/Administrator typically has the two main choices today… And both require at least infrequent attention. There is no such thing as a “zero touch” machine for years unless the machine is very simple and is “rolling.”
Either a standard release which involves major headaches about every 1 1/2 - 2 years
or
A rolling release which incrementally upgrades every few days or weeks. The hope is that if changes are frequent, breakage is hopefully small enough to fix when it happens.
The release schedule can be frozen or otherwise over-ridden but at the potential cost of delaying the pain (And yes, every now and then like any other distro we have Forum questions how to maintain or upgrade a machine that hasn’t been touched for 5-6 years)
Those who decide to try openSUSE/SLES will find that the ecosystem is nearly a decade ahead of other distros in determining what works, and ironing out the bugs.
2G – GSM – development began around 1988 – the first GSM networks began rolling out during 1992 – I was a member of a development team involved with a Base Station Controller (BSC).
3G – UMTS – around about 2005, I was a member of the development team involved with the deployment of UMTS Base Stations for AT&T Wireless – produce patents or die …
Our American colleagues were totally impressed when we took them for a drive test on a nearby German Autobahn – telephone conversations and Internet surfing while sitting in a car being driven at 200 km/h …
A well-known antenna supplier quipped that, “no mast is necessary – simply bundle all the feeder cables together … ” >:)
Yes, most of the hardware is fairly robust but, spare parts is quite an issue – whether or not the semiconductors used are still being produced or, not – is always a moot point …
Just because the capital cost has been recovered, doesn’t mean that the thing will be replaced – there are other costs to consider:
The 24x7 electricity bill.
The 24x7 landline costs.
The 24x7 leasing costs for the area where the Base Station and mast are located.
The financing of the frequency licenses.
The effects of weather on the mast and the antennas …
The staff costs to keep the network tuned and, to continually analyse the traffic being carried …
Actually, for telephony, the available bandwidth allows hundreds of conversations to occur within a relatively compact geographical area …
Never understood why anybody would want to install CentOS, anyway. Production servers in data center are RHEL with satellite and everything. if you want to have a devtest VM, you can get RH developer account and run the real thing. For docker/podman images there is now ubi. So not even there the need to build stuff on top of Cent. And for desktops there is Fedora.
