Slashdot Mirror
Red Hat Announces Product EOL Calendar
BrunoC writes "Looks like Red Hat is getting a little Microsoftish and is quietly introducing its brand new 12-month-only Errata. Quoting The Reg: 'Red Hat's current death list EOLs RH 7.1-8.0 at the end of this year, while 6.2 and 7.0 get theirs as of the end of March.' You can read the whole article here." I don't see how this is "Microsoftish" -- the code Red Hat creates or includes is still GPL, and you can pay anyone willing to fix it. They're not required to support it forever :)
They are a company afterall. You can't expect them to support all their products for an indefinite amount of time. They would go bankrupt!
seeing that i really enjoy using the most archaic versions of redhat i can get my dirty little hands on. i mean, i see how they are of some use, but i dont understand while people are getting antsy and making m$ related accusations...
xao
xao
http://TheHillforum.hopto.org
isn't this old news? I could swear I read about something like this in their "Under the Brim" newsletter at least a month ago.... oh, well. (shrugs)
C|N>K
Pardon me, but if using short product support policies is "microsoftish", then Redhat is more microsoftish than Microsoft itself. Last time I checked, support for Windows 95 was dropped on December 31 and support for Win2000 will be dropped in 2008. That's 8 YEARS per product with a possibility of extening support for corporate customers.
I do not recall Redhat supporting any of their distro releases for 8 years.
noob - "I am having a problem with USB..."
RH person - "What version are you using?"
noob - "Uhh... version 5.0 I think..."
RH person - "FUCK OFF AND UPDATE YOUR SHIT MAN!!! IT IS FREE!!!"
It isn't like anyone was paying for their distros anyway.
The whole concept behind Open Source is that selling service is the way to make money. However, when no one is paying you and demanding your services even still, there's got to come a point where you realize that your "customers" are simply taking advantage of you.
Bravo, Redhat. For finally realizing that money doesn't come from beggars. Now maybe my RHAT shares will be a shit.
I have been pwned because my
Here is an example of the rapid advancement expected when utilizing open source development. Proprietary users will think "Retiring a major OS in just a year? That's crazy" - while we Linux users have grown accustomed to such things.
Um, seriously. End-of-lifing a product is just a plain good idea, whether you're talking about open source, closed source, or something that isn't even computer software. In the real world, it costs way too much to keep a support infrastructure in place for a product that is only being used by a small amount of the population due to its having become "obsolete" (even if only as a marketing matter). While it sucks to be one of the people who still uses the product and doesn't want to upgrade, there's really no alternative but to cut people off eventually.
One of the virtues of free software is its rapid development/update cycle. Why would should a company based on this development model sell software as if it were never updated?
I don't see how this is "Microsoftish"
Maybe that's because you don't have to admin anything important. An annual upgrade treadmill is a huge burden on IT staffs that have to prototype and test rollouts for upgrades. There is a reasonable support timeframe between zero and indefinite and one year is not it.
No one pays? My employer shells out a few grand a year for enterprise RHN...
Microsoft also still makes available all the online information (MSKB) for the discontinued OS's.
And its a bit different with redhat, most of the components of old redhat releases are current projects still making releases, theres nothing stopping you from doing the upgrade yourself, chances are if you cant you can prob get your slightly geekier friend to compile and rpm it for you. With windows, think someone is still working on the windows 3.1 version of the networking components?
Jesus saves, everyone else takes full damage from the fireball.
"I don't see how this is "Microsoftish" -- the code Red Hat creates or includes is still GPL, and you can pay anyone willing to fix it. They're not required to support it forever :)"
I think the more important question is. Why is everyone so gung ho about seeing every RH action as "Microsoftish"? As many have already argued RH couldn't be another Microsoft. Has Microsoft scared us all so bad that we jump at the slightest movement by a commercial company? What about all the other commercial companies out there? Aren't they doing something "Microsoftish", or is it just RedHat?
Since three years warranty on server hardware seems to be not uncommon, possibly this is the thin air Redhat seem to have plucked this number from?
It's nice to know that when you get your shiny new 8-way Xeon with untold amounts of RAM you'll be able to leave it in production for the span of its warranty without having to worry about re-installing due to the OS release on it being EOL'ed.
Where this falls down is twofold: 1) servers are still useful well past three years, whether they're warrantied or not, and 2) some vendors for extra money will extend warranties up to five or so years (my employer has started buying Dell boxes with five year warranties pretty much as standard).
Come on, Timothy, that was cheap :-) Of course it's "Microsoft-ish" because it forces companies who want support to upgrade. Yeah, sure, you still have the source code, but in a company that doesn't mean anything if you're not getting support. Half the reason why Red Hat is so popular (over the "free beer" Linuxes like Debian) is because when a company puts it on their systems, they can be assured of getting professional support. This is really important for the PHBs of the world - they don't want to hire some in-house hacker with tattoos and spikey hair to "support" their installation.
Of course, even though it is Microsoft-ish, i don't think that's a bad thing. Forcing your clients to upgrade is better all round - it's better for the economy because it's creating sales which lead to more R&D spending, plus you can ensure your clients are running the latest version which should cut down on the bugginess or flakiness of their software. If Microsoft had had a more aggressive "push upgrades onto the client" scheme, all the internet problems we saw last week wouldn't've happened, because everyone would've been running patched SQL Servers anyway.
I got a sig so you would remember me.
Maybe the word 'quietly' is what's microsoftish. But actually Microsoft is quite vocal about end-of-life announcements hoping to spur new sales of the latest product suites. Actually, the poster really should reference Oracle, whom is the master of desupport notices; often on the order of 'this product will self-destruct in ten..nine..'.
I guess Red Hat is being microsoftish by trying to make a profit (maybe someday), or trying to keep the majority of it's users somewhere in the middle of the bell-curve (you spend 90% of your time supporting 10% of your users who refuse to upgrade), or maybe it's the windowsupdate.com like ability to patch over the web.
I think they're more Microsoftish than you may think, and I say 'right on!'.
"This isn't a study in computer science, its a study in human behavior"
One of the major reasons to choose RedHat is their reliability. Thouroughly tested software you can rely to have on your server for at least a year without having to worry about it except for bug- and securityfixes.
Ok, 7.1 is rather old, but discontinuing support for 8.0?
IMO professional distros should always support their latest, and their last major release, so in RedHats case 8.x and 7.3, and not drop support for 7.3 until 9.0 is out.
After all, support is, like, the thing theiy make money in the first place!
my
The thing that comes to mind was the discussion the BMW exec had with a number of attendess at a tech conference. He point out that they are required to support cars with parts, etc for Ten Years. And the obvious question was how may people there were running things that were ten yerars old, nevermind able to get support for it.
Now we get to End of Life issues. How long should software be supported? Ten years for something like software, Is this even reasonable? It's important for the embedded market, at least.
"It is a greater offense to steal men's labor, than their clothes"
I think that one of the major beefs against Microsoft is that they require you to PAY to upgrade to the latest version. I don't see that dropping errata support for something that will cost you a grand total of $0 (if you have fast net access) or a few bucks to get new discs from one of the cheapbytes-type places out there.
Personally, I'd rather see them drop the old support in favor of providing a higher level of service to the paying customers. (This isn't a dig on their service, which I think is great - we're paying customers at work, and RHN is a tremendous tool.)
This ISN'T microsoft-ish. Microsoft atleast supports their products for a little while. What this is, is a company screwing the living hell out of the community that's supported it. I've sat here and sold personal version after personal version alot with errata accounts to clients because a) cheaper b) would be supported for quite some time with good security updates and wouldn't always require upgrades to continue to use their other products.
/. apparently didn't like). I've started installing Gentoo on my workstations here already and within the next 4 weeks my redhat boxes will be gone as well.
Now I have to turn around and tell them that Redhat changed it's game plan and convert each one of these clients over, or let them continue to pay me to constantly upgrade their network just to keep them within their errata entitlements. I for one....basically said to hell with redhat about 5 hours ago (incidently right after I submitted my story that
Face it people, the people like "us" have made redhat and they just turned their back on us for the corperate world.
Don't get me wrong, I have NO problem with end of life, but 1 year for what's there now. The woman I spoke with at Redhat (yes I did research it directly with the company not just reading what nimrods say) she said that after this first round, there's going to be another change. Anyone using personal or the "free" version (and probably the professional) will ONLY be eligable for errata during the time that the release they are using is current. As soon as they release another version, errata for the older is gone. In other words, since redhat releases usually twice a year....that would me 2 upgrades a year just to keep yourself up2date. Screw that.
It becomes microsoftish when an upgrade is not a free download away.
In SOVIET RUSSIA... erm...NSA AMERICA, the Internet logs onto YOU!
NT4 will still be supported for another year... that's about as long as Red Hat's supporting 8.0 (give or take a few months).... Windows 95 has been out for... what.... 7-8 years? As another poster pointed out, if this is Microsoftish than Red Hat is moreso than Microsoft itself.
I've never bought support from Red Hat... I just d/l the discs off a mirror and go (and am building my own system in my spare time, though I need a working system in the meantime). Anything Red Hat releases you can get for free and, after all, Linux is still a vastly growing community. For RH to EOL their products in such a (comparatively) short time-frame is expected, given the rapid growth of the core Linux software.
If the distributions were dramatically different between versions on the inside, I could see why someone would be worried about a short RH EOL. But they're at the mercy of the defacto standard version (if you could call it that) of the thousands of core packages that make up the entire distribution. Everyone's trying to jump on the prelinking bandwagon offered with glibc 2.3.1 (which is a killer for a graceful upgrade), but the prelinking feature is about as far from backwards compatible as you can get.
That would make sense except for the fact that it invalidates the very reason for RH's existence, which is to provide precisely that sort of service. People who buy RH are the ones least likely to maintain the code.
"I don't see how this is "Microsoftish" -- the code Red Hat creates or includes is still GPL, and you can pay anyone willing to fix it. They're not required to support it forever :)"
But isn't that what you're paying RedHat for when you buy support from RedHat? By cutting their support, they're cutting the one service that paying customers actually want (unless they buy the software as a donation). MS just patched NT4, which has been out since '95 or so, and you're criticizing MS and excusing RedHat. Give me a break.
Vote for Pedro
Yeah, as the product responsible for Linux I can sure see myself explaining this to my boss (who is very pro-free software): Er, yeah mate. We just hire a bunch of hippies if Red Hat support runs out on the server products we run. I'm sure Oracle will be more then happy to support our home modified kernel sources. Sure a great career move on my side.
Sorry, this is just plain dumb and makes me wonder if Red Hat indeed is a good choice for this company. We are talking of a major divison of one of the biggest logistics companies worldwide.
A one year time frame is just plain unacceptable in a corporate environment.
I think it very much depends how Red Hat handles this on their enterprise level support contracts.
(I read the part about the three year life cycle for their "advanced server" products. Which ,imo are just a scam in the first place).
ich bin der musikant
mit taschenrechner in der hand
kraftwerk
Different things.
MS requiring an upgrade is forced because no one else can support or fix bugs or security issues on old Windows or MS-DOS versions.
Red Hat is just stopping their own support for old versions, but anyone else can fix their bugs or security issues, and support it, because they have the source code to it.
No one's forcing an upgrade on Red Hat's half.
Human nature is the same everywhere; the modes only are different. -- Earl of Chesterfield
I agree with your post. Honestly, I don't generally like much that MS does on the business side, but product support is one that seems _fairly_ reasonable (at least in the OS realm). Really, '95 support just expired...that's a long time to support a piece of junk.
.spec file. Upgrade yourself, or even better, setup your own local RHN server to roll them out for you. You don't have to patch things like Evolution (not installed on your servers anyway, right??) on a server, just the security related errata!
I also don't think that a year is unreasonable for RedHat to support an OS for...especially considering we're talking about Open Source products here. Really, RHN/up2date is nice, but not a necessary component.
There are two things to consider here.
1) Home users: who cares if they have to D/L a new ISO every 12 months...sure, it'll cut into the pr0n allowance, but no biggie (sorry dial-up users, you'll have to shell out $5+shipping for a disc).
2) Corporate users: upgrading servers is a pain. It's done as little as possible. Open Source is great in this situation. Upgrade on a package by package basis. It's fairly easy to build an RPM...especially when a lot of projects include the
And to top it off, corporations should be using Advanced Server anyway, or have the $$ to pay RedHat for some on-the-side support deal...this happens all the time.
Even a non-RedHat supported RedHat is still a very maintainable system.
-Ben
Why not tie the EOL for redhat desktop products to the vailidity of of the RHCE for it? For corporates, its going to be very hard to get approval for certification training if you know the EOL of the product is less than the period of the certification (I recall the figure of 2 major releases being mentioned by the instructor). This could damage RedHat's rep in the training market (one of their key publicity points in the last few years).
I'm surprised that people are still running RH6.0. It's far less secure than 7.x or 8.0. The desktop (and server environment) are much better as well. Sure there are some libc5 legacy apps but there's really no excuse for a server to be running it. Upgrade or do a fresh install and use the newer features (like journalling, LVM, iptables, 2.4 series kernels etc) because they make an immense difference. RH7.2 really should be a minimum if you are serious.
Actually, they're cutting services for the only people who do actually pay for something. Very bad for business.
Vote for Pedro
and you can pay anyone willing to fix it. They're not required to support it forever :)
Since the code is otherwise free, service is the *only* thing you're paying for - it should be top notch.
http://www.redhat.com/licenses/rhlaws_ita_us.html? location=United+States&
especially the report and auditing section.
I support a lot of redhat machines.I appreciate that they ahve to make money and all, but really. I don't call them and ask for support. i don't use the RHN. i run a mirror.
I did call once and ask for support...I got tossed back to HP!
-- Who is the bigger fool? The fool or the fool who follows him? --
Way back in December LWN covered this and I think Alan Cox voiced his thought that people (not RedHat) may try and make a business out of support 6.2. Now there's an idea...
I mean, even Linus himself end-of-lifed the fantastic 1.2.13 kernel a long time ago.
..and is quietly introducing its brand new 12-month-only Errata.
"Quietly??" Hardly. I've received notice of this at least three times in the past couple of months from various RH newsletters. I even considered writing to let them know I had gotten the message. It's been on their errata web page for over a month (at least since 8.0 has been out).
I guess this shows you CAN'T count on people to get the message unless you beat it into them, or perhaps this whole article is a RH troll to actually get the message out??
I now expect to receive several other explanatory e-mails from RH after this slashdot article.
RH8.1 isn't even out and they are announcing an EoL for 8.0? What kind of crack are these guys smoking?
It would be awfully nice if they would have something to upgrade TO before planning the end of support for the current line.
Considering that the "autoupdate" stuff in RH doesn't work very well, most upgrades require either backing out current data and rebuilding, or making a new server and moving the data over. This requires hardware and expensive man-hours. So much for that low total-cost-of-ownership that linux is supposed to provide.
You know, you are SO wrong. Ofc. the beggars(Students, the primary slashdot crowd, etc.) will not mind they now have to upgrade to get supported versions, they do that each other week because they don't have anything else meaningfull to do.
But, in the real world, you just don't upgrade each week. First of all, you don't have the time to do it, second many of your services are so complicated they might break seriously if a patch is applied, and ofc. all of this has to be done on a working live system in a very narrow timeframe, which leaves you very little time for errors.
This is a very poor move, now that Linux has been accepted in the business world. This will clearly throw some people back to Windows, because their lifetime is bigger, and the systems are easier to update. Not that I don't know how to use the patch command, but hey, most people would like just to double-click on SP3 and then wait until it is done.
-H
It's not like they're killing you because you need to pay $$$$ to upgrade to RedHat 2005. All you need to do is download the new version and upgrade. I do think a year is kind of short ... Like someone said maybe they should support at least the last mayor release.
... 4.7? 4-stable? I don't see them recommending 4.4 to anybody right now, even though it was pretty stable. 4-stable is the recommended Server software until what ... 5.2?
I mean, how does FreeBSD do it? Do "they" fully support 4.0?
In this case, the question is, how long do you give support for a product so that your costumers feel that the upgrade will be stable, after all, it's not like they _have_to_ request PHB for a few grand to upgrade.
All I can say is good for RH, but I would have thought a "previous major version support" would have been better.
Then again, you can always just get the latest stable version of whatever software you need so much (apache? sendmail? vi?). And in the case you're building a new system, why not install the latest stable distro version anyway.
I do understand that people have different requirements, so there are exceptions to everything. I mean, you can always do your own RedHat 5.0 support department if you wish.
cl
Reply . . . let's get it over with.
But the RHCE program is geared towards this same "consumer" release. Current RHCE is for Redhat 8.x version and you have to get recertified every other (consumer) major release number. So, what good is RHCE? You get certified to run your home Linux box then?
I know some sites are still running Solaris 5.2 (which was de-emphasised about 5 years ago). It takes some companies almost a year to get their software really stable. Forcing them to replace their OS on a yearly basis is going to discourage movement to redhat
From a marketing (as well as technical) point of view, theis seem s like a really bad idea(tm).
OS Software is like love: The best way to make it grow is to give it away.
I entirely agree. I find it fascinating that while Microsoft, the bane of computing everywhere, is booed for end-of-lifing Windows 95, Windows 98, and the like a good seven or eight years after they are first shipped to the masses, most of the replies here (at least so far), are condoning Red Hat's end-of-life policy. Talk about viewing the computing world with rose-tinted glasses. I suppose what isn't good for the goose is perfectly fine for the gander.
"We are all in the gutter, but some of us are looking at the stars." - Oscar Wilde
Actually this may explain why RedHat is doing things this way. If major apps require old versions of RedHat then that would, all by itself, force a lot of people to refrain from upgrading - even if they would like to. Even worse if different apps require different versions then that is going to seriously undermine the value of the RedHat Distro.
By limiting support to a small number of versions they encourage vendors to keep up with more recent releases of Redhat, and make it more likely that all the major apps will actually work on one of those supported versions.
Since Redhat doesnt make loads of money exactly from keeping out of date systems alive they havent go any incentive to keep those old systems running. Looking at statistics would probably show that very few run older versions of redhat. You can still use the PRO versions and get longer lifecycles if you run a business. For redhat to keep systems released every year up at no cost isnt good business and we dont want them to go titsup do we?
The incredibly fast development of linux right now is making older versions obsolete very fast. If you want to run something really old you should use debians stable version since it is rock solid and dont tread on the edge like most other distros right now.
HTTP/1.1 400
I have one. Windows XP.
I don't have to pay $800/year for Windows XP for every desktop install of it. It seems my choice for Redhat on the desktop is either AS at $800/year for three years of support, or the "consumer" version for a one year support cycle.
You state that they are the MS of Linux like someone came down from on high and pronounced it for all to see. Last I checked, I couldn't:
1) download an absolutely free (as in speech, as in beer) operating system
2) sign up for automatic update-checking (without signing a privacy-undermining EULA)
3) download ANY software that goes into said OS, often in multiple versions, including cvs and beta
4) get all the code to any of the above to hack on if I want to
with Microsoft.
Just because they are the largest GNU/Linux company, and the most widely-known, doesn't equate them a priori with Microsoft.
B
ps-- That Aaa aaa BBB thing makes more sense to me... All A's should come before any B's.
"We must still have chaos within in order to be able to give birth to a dancing star." --Friedrich Nietzsche
I think you missed the main thrust of the point that was being made.
If Microsoft stops supporting something then it will never be supported by anyone. No bug fixes, nothing. If RedHat stops supporting something then there is nothing to stop you from paying someone to pick up where RedHat left off. Hell someone might even find a viable bisiness model here by servicing old RedHat distros (OldHat?)
Also, cut it out with the "M$" crap.
Microsoft built its early business on porting its BASIC programming language interpreter to several 8-bit microcomputer platforms and licensing it to the computer manufacturers. In line-numbered BASIC, the name of a variable of type string ends in '$'. A valid program in "Applesoft BASIC" (the BASIC interpreter in the Apple II ROM, developed by Microsoft):
I find using a BASIC expression to refer to a BASIC vendor just as valid as using the pattern *n?x to refer to a family of operating systems whose shells recognize the name of the operating system in that glob pattern.
Will I retire or break 10K?
That would make sense except for the fact that it invalidates the very reason for RH's existence, which is to provide precisely that sort of service. People who buy RH are the ones least likely to maintain the code.
You missed the "or pay anyone you like to maintain it" part.
It turns out (I'm certain) that if you REALLY wanted M$ to support '95 for you, they would. You'd just have to fund the entire effort yourself, plus whatever profit they wanted.
Likewise for Redhat. However, there are a lot of other folks you could convince to support whatever old version you want in the case of Redhat. It's not really feasible to try to get someone else to support M$ software - not the way you'd like, anyway.
I actually worked for a company where this was done. We were paid VERY good money to support a free (as in speech) product. This back in '96 or so.
I've still got one box purring away with 5.2, another running 7.0 and the rest of my boxen at 7.3. All happy clams running what they're capable of with no issues. I can still build packages from source or snag available RPMs for any critical security or bug fixes. It's not like Red Hat's preventing me from doing that. They're just trying to manage their customer support model and remain profitable. Good for them!
RedHat was founded in 1994.
"Can of worms? The can is open... the worms are everywhere."
Perhaps in the real world of beanie-wearing community college graduates, yes. But in the real corporate world, there are testbed servers on which to test upgrades, patches, etc. before rolling them onto the production servers. Often times there are also redundant servers which can be taken down, upgraded, tested, and put live one at a time.
Further - there's no requirement to upgrade once/week, but at the very least keep packages less than one year stale. The Internet as a whole got a kick in the goodies this past weekend by sysadmins who wouldn't patch a software vulnerability that was more than 7 months old (and by the network admins who allowed access to the servers via the public Internet, but I digress).
If you haven't upgraded your Linux systems in 6-12 months, I'd love for you to send me your IP address(es), because I'd like to send you a few packets pertaining to;
Out in the "real world", systems administrators apply patches, fixes, and upgrades to their software regularly to avoid being used as a staging ground for one of our recent many DDoS attacks, or having their corporate data stolen.
It's the lazy, incompetent, certifications-are-king sysadmins out there who give us a bad name. They're the ones who adopt the theory that applying updates is "too hard", and claim that "things could break" which they use to justify their ignorance of best-practises security.
If your company's assets are riding on IT software and you're having trouble keeping up-to-date, talk to your vendor and ask for help. Have them justify the money you fork over to them every year and do something for you. If RedHat is your vendor, ask them for assistance in migrating your server farm from 6.0 to 8.0. If they won't give it to you, inform them that you'll find another vendor, and that you won't be spending $30k on another support contract. If you've already spent it, contact your lawyer.
"Real World" does not, nor should it ever be confused with or used to justify laziness, ignorance, or apathy. It's thinking like that that got us into our present state of dissaray.
BD Phone Home!
Shameless plug. Like you weren't expecting it.
This new policy dramatically increases the cost of maintaing Red Hat Linux in any business computing environment. I have RH machines with uptimes longer than 12 months.
Given that Red Hat has 6 month release cycles, a lot of people are going to find themselves with software that is to be EOL'd six months after it was installed, when at the time of installation it was the latest and greatest version.
I'm very partial to Red Hat Linux, but this new policy makes me rethink my opinion.
What has *science* done?!? -- Dr. Weird (ATHF)
That said, Linux has an extreme level of upgradability. Using Red Hat specifically, I ran version 5.1 and upgraded it using newer and newer packages and custom kernels. The result, before I decided to restart from scratch, was mostly based on RH 7.
Even a kernel update -- custom or packaged -- usually does not require user level software changes. When it does, the updates are usually backward compatable so you have a fall back option. This means that if someone runs RH 6, and a local exploit or bug is found in the kernel or other software, they can update to a version that will not have the hole.
Is upgading single packages painless? Not necessarily, though the painful parts are usually because of package dependencies with non-critical programs. Having a mix of packages from different 'versions' is entirely possible as long as you handle the upgrades in a conservative manner; update only what is necessary not every package on the system.
A firewall can not protect you from yourself. Turn off what you do not need. Do not use the firewall to do your work.
I thought this was supposed to be a pro-linux site?
Timothy completely missed one of the key points - that the "advanced server" version has a 3 year support plan. Its still not great, but to have "Redhat announces 12 month only errata" (or words to that effect on the front page of Slashdot is just going to give Redhat a bad name with the casual observer. We all know how few people actually RTFA...
smash.
I run: Windows, OS X, Linux, FreeBSD. Just because you have a hammer, doesn't mean everything is a nail.
I suspect that's why they are releasing Advanced Workstation
It's suprising that some folks are just now finding this out. I noticed this a while back while trying to get a decent fix for the fubar xinetd package Redhat is pushing on its pre-8.0 distros.
I was really suprised by this since a long lifespan is the one thing that RedHat had over Mandrake (Mandrake's product lifespan is 2 years from date of initial release..) I don't know about the rest of you, but I have servers running right now with 2 years of uptime..some are in the same city as me, some are colocated in other cities. I can't upgrade these systems without either flying to the colocation site or having them mailed to me.
I came to precisely the same conclusion as the folks in this article. If you're using Linux on a server, it's stupid to use anything other than Debian. The commercial distros NEED you to upgrade, whether or not there are any compelling new features in their new versions. The Debian developers could care less about you buying a new set of CDs every six months.
It's pretty funny that RedHat seems to be following right in Mandrake's footsteps here. It will be a great boon for virus writers if they really do drop support for all those 7.2 installs out there...but I can't imagine that serious sysadmins will put up with this for very long.
Real businesses, with thousands of servers, can not upgrade every year. Besides the actual time to do the work of upgrading, there is testing that must be done when you have real money at stake, downtime caused by the upgrade, etc.
I work for a real company. I can't use an unsupported operating system. I can't upgrade every machine every year. I can't even upgrade to the latest and greatest (e.g. RedHat 8 and Solaris 9 are out of the question), because it is too untested. These are the business realities, not factors that I or any other individual have control over. A single incident (e.g. a server crashes and whatever sort of failover is in place does not work) can cost more money than my yearly salary. A single hiccup (e.g. a 1 second network outage for a single machine) can cost more than my paycheck.
Sun at least makes guarantees that binaries that worked on previous versions of Solaris will work on new versions. (If they pass a test suite). RedHat makes no such guarantee.
I thought I was making real progress to replacing Solaris servers with Linux servers. But with this announcement, I don't know what to do. If I deploy RedHat, I am adding a substantial (and mostly hidden) cost and risk. RedHat seemed like the logical choice, but my next course of action is going to be to investigate alternate supported Linux distros (IBM, Sun).
Red hat is doing this because they don't have the resources to support 6 different versions of their O/S full time. These are just errata updates. If having the errata (security, bug fixes, performance issues) is important enough to you that you're going to care, then you can:
(1) Download the latest RH and buy RH support for it (because this is really just about support),
(2) Hire your own admin to stay abreast of these issues, or do it yourself. (But this is why people pay for RH service, so this isn't a likely option, or
(3) Contract out to another company to step in and do errata updates on EOL'd RH distributions. It's legal, but probably expensive, unless a bunch of people band together to do it.
Anyways, as the argument goes for contributing back to the linux codebase . . . It doesn't make any sense for a company to update programs without trying to get their updates put back into the codebase. Same thing is going on here. It doesn't make sense for RH to back-patch all of these older programs just for errata updates. Too much time and effort that could be spent on the real task: creating more advanced and better working programs. You don't see many people doing updates to Linux Kernel 2.0.39 these days, do you? Wonder why that might be. . .
Karma: Not Particularly Funny.
Man, this stinks. I read it in Red Hat's newsletter but forgot about it until today.
.. now we have to upgrade these production servers? Can I do that over the network? Am I supposed to talk them through it over the phone? It's going to be a huge deal to upgrade all these machines.
Like others here I have convinced several clients to switch their servers to Red Hat Linux. The clients pay us a monthly fee to read the nightly reports as well as pay the cost of a Red Hat Network account. Keeping these machines up to date and secure is a total breeze and MUCH cheaper than the proprietary alternatives, for all involved.
I also chose Red Hat over my usual choice of FreeBSD because of the ease of upgrading (i.e., when a bug comes out, just run up2date, and everything works because Red Hat back-ports patches to keep versions stable).
These clients all have 7.x, including 7.0
And you can bet I'll be upgrading them to FreeBSD (with Linux compat as needed), where upgrades can be done from CVS. Sure, not quite as nice as up2date, and version problems will probably come up, but maybe someone's written something to alleviate the problem.
I like the idea of them supporting the last two major releases (7.x and 8.x right now). If I have to pay more for Red Hat Network support on an older version, I'll do it. But it will be quite a chore to upgrade all these machines to 8.x by the end of this year.
I hope Red Hat comes up with an easy way to upgrade remotely! Otherwise, bye bye Red Hat, and I have to explain to my clients that all that stuff I told them about Red Hat saving everybody money is a bunch of BS.
But it seems like a really stupid move to me.
Isn't RedHat trying to woo big corporate/enterprise accounts? From what I know of the corp/enterprise attitude (admittedly not a lot), they don't wnat to have to upgrade the whole OS on a yearly basis in order to stay up to date.
I do realize that the packages themselves will very likely be upgraded and that any admin can go get them and apply the updates himself, but isn't up2date and its associated collection of updates in one easy to find place one of the biggest selling points?
How is RedHat not shooting itself in the foot on this one? Someone please explain it to me, I'd really like to know.
Information doesn't want to be anthropomorphized anymore.
You missed one category:
3) Educational users: upgrading our servers is also a pain. We do it as little as possible. Our main server is still running 6.2. I am in the process of replacing it with an 8.0 server, but it takes a lot of time. We cannot afford Advanced Server. We are going to be left out in the cold.
I do support Red Hat however I can. I buy RH Pro for each release. That is about the limit of our budget.
You can't? Just guessing here, but I bet you use linux in your dormroom or your bedroom.
In April, a co-worker and I upgraded roughly 15-20 machines from RH 6.1 to RH 7.2. We don't have IT staff as such, we are both scientists in academia who happen to know a fair bit about system admin. We work in an academic environment.
Our 15-20 machines are all slightly different. They all needed to have certain config files backed up and restored. They each have a different person with different skills and different requirements sitting in front of them. So, how did it go?
The first machine probably took 2 hours of fairly close attention to install. Everything had to be documented so we could reproduce it. Then, I used that machine for a few weeks, noting what else needed to be tweaked and installed. Then, one by one, we installed the OS on the other machines. This process took about 2-3 weeks and took say 30-60 minutes of real work per machine including the updates. Then the users got a hold of them. One person notices that program X doesn't exist anymore. Another notices Y doesn't exist. Someone else notices that xvscan doesn't work, so we have to figure out how to use xsane. This continues for a month or more, each time requiring one of us to install more software on the systems and test it out. The first machine (mine) is determined to be completely out of sync due to different choices in the installer, so it is done from scratch.
I highly doubt this is a "rare case." These are just desktop machines, not even mission critical servers (although one was a web/db server).
I sure don't look forward to repeating this excercise in January 2004. If Red Hat's options are a 12-month upgrade cycle or $800/machine, we'll find some other company. But, their promised corporate desktop may be the answer for us if its priced reasonably.
I first noticed that change when installing Solaris 2.6 a few years ago. Turned out it was the locale setting that defined the sort behavior; the default locale used to be "C", which corresponded to plain ASCII, and the new default was "en_US", which gives the behavior you describe.
To restore the old behavior, on Solaris and on Linux, you can set the environment variable "LC_COLLATE": export LC_COLLATE=C in your .bashrc (or .bash_profile, or /etc/profile).
I kinda see your point, but this isn't really the same thing. There will always be perfect, consumer level, support for products like Linux. It's grown without the help of companies, and will continue to grow- with or without them.
I am by no means a fan of RedHat (Slack for me), but I think that they have a legit and smart model here. You're essentially buying support when you buy their (reasonably priced) distribution. If you don't like it, then don't pay them. The ISOs will still be there tomorrow. And even then, the source code is there for the consumers to improve upon.
It isn't quite the same thing as Microsoft's forcing users to upgrade closed-source products with no new features (Office comes to mind), but even MS has to make a buck. Weather or not I agree with the means by which they do so is another story.
Overall, I don't feel that it is fair of The Register to present things this way. They've lost a lot of respect from me.
Linux distributions are a collection of three primary factors, and a number of secondary. The primary factors are;
Emphasis on #2. Like most other distributions, RedHat maintains their own custom packages for their software applications. That's why each distribution will in-turn address security updates on the likes of BUGTRAQ; each of them is distributing their own RPMs/DEBs/TGZz/etc. to fix the problem. That's what RedHat is ceasing; support for their custom software RPMs.
You'll always be able to download/install the latest kernel, be it via RPM, SRPM, or sources from kernel.org; I don't believe that's an issue in the slightest.
Once you've upgraded all the packages to their current level, all that's left are the few packages (Base layout, package management interface, etc.) that comprise the "OS". If you require support but RedHat declines it on this basis, simply apply the updates and call them back.
The high school I worked for had testbed systems. They're just common-sense. When you're deploying mission-critical, or even just important updates to a system, it's always adviseable to test it first. Be it a server upgrade or a workstation image, it's just good business practise to test it before you put it live.
The advantages to spending the money preemptively and having a solid software update policy in effect will save the company countless dollars, man-hours and heartache when and if that policy would have helped the most - in the case of an intrustion or major software meltdown.
BTW - Even when RedHat christens a newly released RPM that upgrades an existing package, it should be standard policy to test this upgrade on a testbed before putting it live, so the extended product life cycle won't do you any good anyways; damage done is money lost.
A testbed server doesn't have to be some kind of elaborate, highly-involved process. Essentially, you keep a 'good' image of the install base, else simply a mirror of the production server. Install this on a similar hardware platform (ideally a machine with the same hardware configuration), apply the upgrade(s), and test the robustness of the new system. Re-boot it and ensure sofware comes up. Watch for panics. Put some load on it and make sure the services don't fall over. (This, of course, will be done on a case-by-case basis, as all servers face different load scenarios). That done, migrate the changes to the production server, monitor it closely for a few days (even if you just watch Big Brother or a couple of log tails on your desktop), and call it a day.
I maintained upwards of ten separate workstation images and five servers using this methodology and virtually never had problems with any production setups as a result. We continued to stun the IBM tech reps (with whom we had a support contract, but that's a story for another day) when they'd come to the school and find themselves with absolutely nothing to do. They'd start to ask us about a problem they'd seen at other schools and we'd stop them with "Fixed that three weeks ago before it became a problem."
BD Phone Home!
Shameless plug. Like you weren't expecting it.
It's more than that, though. Microsoft forces "upgrades" by creating software that requires the new operating system. I don't see Redhat doing this. When new software is released that requires new libs, those can be installed separately without needing "support" from Redhat.
Microsoft has to extend their product support because at 3 years they have usually fixed enough bugs that companies can start to use the software. If they promised not to fix any more bugs, no one would ever install their software.
But DNS servers and print spoolers that have already been running for several years do not need support from RH. In the rare case that an upgrade is needed (for a good business reason such as new abilities or a security fix,) most linux administrators can easily handle it.
This announcement means two things to me:
1. RH will no longer maintain the lists of fixes/upgrades for me unless I download/buy a recent version.
2. RH will no longer produce and test binary installs for me. (Most of the software we use is available as RPMs from the developers anyway.)
Oh well. I may have to stay current by checking ten websites instead of one.
I spend my life entertaining my brain.
If you're trying to purchase a few dozen (much less a hundred or a thousand) desktop machines for corporate rollout, it's going to take you a few months to get the budget approved. Then you spend a month or two on the RFQ and RFPs, another month or so going through them, and another month or so finalizing the decision. Add on the order time, receiving time, and software installation/configuration time, and you're hitting 9-12 months before they're even hitting a user's desktop.
So you've got a good chance that by the time your users first turn on a RedHat desktop, the support has been dropped.
Congratulations, RedHat, you just knocked yourself out of competition for the corporate desktop. With Mandrake dead, that leaves SuSE as the only real contender for a corporate solution on the desktop.
On the server side, consider that it typically takes at least a year for third-party vendors to certify a distro as "supported" for their products. Sometimes it even matters -- Sybase 12.5 would only run on a certain patch level of RedHat 7.1 last time I tried it (Mandrake 8.1, 8.2, and SuSE 8.0 could not even prepare the storage space for the database without crashing, much less run a server.)
I know that most corps are going to have special contracts set up for support, but that doesn't help those of us on the development or consulting side of things who don't have the budget to pay for full AS licenses just to get a system that doesn't need to be rebuilt annually.
If I want to rebuild systems annually, I'll go back to Microsoft-based development -- there's more work supporting that junk anyhow.
I do buy full distros to support the vendors -- and end up spending far more on Linux distros per year than I ever did on Microsoft products as a result. I have RH 5.2, 6.2, 7.0, 7.1, Mandrake 7.2, 8.0, 8.1, SuSE 8.0 and 8.1 -- all full box sets at $75-100 each. Even when I don't install them, I buy kits just to help keep the companies I believe in afloat.
I sure don't appreciate RH trying to rip me off as payback. Even with RH normal pricing, who in their right mind is going to pay $150 for a full current release of RH, for which you only get a few months update support, vs. buying a generic copy of the disks for $20 plus shipping and paying less than $150 for a full year of RH update support? Such nonsense would be why RH 7.1 was the last distro of theirs I bought or installed -- I don't believe in their model anymore.
I do not fail; I succeed at finding out what does not work.
I couldn't believe my eyes when I read the announcement on Dec. 19th when it come though on the redhat-announce mailing list. I had often wondered why RedHat was still supporting 6.2 but I would have never imagined they would ever go to only a 1-year life-span. That is just way too short. I have been using RedHat for my client's servers but now I am looking for another distro to switch to that has some kind of reasonably easy update procedure for patches and such. I'm currently planning on not installing any new servers using RedHat Linux after June 30th of this year due to this policy change. I do not fell it is right to change my clients full price for a new server install when I change them over to whatever distro I finally decide to switch too so I will likely loose some potential income over this. Speaking of which, does anyone have any good suggestions of distros to use for servers? I'm currently looking mostly at just going to Debian and I am also looking into Slackware. I have used Mandrake in the past for servers and did not find it satisfactory for a server (its perfectly fine for a desktop though). Does anyone have any suggestions that I've overlooked? As far as a desktop distro goes, I'm still trying to decide if a 1-year EOL is reasonable or not. It would be nice if RedHat changed the EOL of 7.3, 8.0 and future releases to be closer to 5 years, or even the 3 years that they are planning on doing with their Advanced Server line. The one thing that gives some hope that the 1-year EOL is not in stone is this line from the announcement: "Therefore, starting with Red Hat Linux 8.0, we have updated the errata support policy and will now provide errata support for all releases of the base OS for at least 12 months from the date of the initial release." (emphasis mine)
I've also been wondering why this has taken this long to make it onto Slashdot. I was expecting to see it on the front page the day it was sent to the redhat-announce mailing list or at least within a week, not over a month later.
If it's true that companies want support for their installations so they don't have to upgrade, then I'm betting that it's worthwhile enough that someone could start a little company providing support to old RH installs.
Think about it; you don't have to do NEW development as you might if you were doing a full distro - RedHat is paying for that. The bugs are going to be identified and repaired by others - you just have to patch up the software and send it out. Wouldn't need to be that big of a company, either, I'll bet - half a dozen techies and a few biz types would probably do it.
You're a fool. You've been bamboozled into believing that everything has to be a commercial solution or it's worthless, dead, or unsuccessful. I've said it before, and I'll say it again: (sing along, kids!)
/. are not buying music and movies. thats why we would rather run anything at all but MS operating systems. And that is why we would rather code opensource than closed. The playing field is a place of crossfire and even friendly fire, so no matter what position you take, somebody will believe (not just think, believe) you to be wrong. Trust me.
Communism sucks for governments, but it works great for groups of people with a common goal!
That's right, kids...opensource is good for you. It lets you give your time to the group, or COMMUNE, and give back the results of your productivity to the masses for equal division! and the great thing about software is, that unlike the finite products that software corporations wish it to be, can be copied effortlessly countless times, with little distribution cost to anyone!
So let's all try a nice big glass of Communism today, and stop worrying about whether Capitalism is going to benefit from our pinko operating system!
but seriously, this whole copyright and software thing is just like the cold war all over again, except this time everyone who has actually researched their stuff realizes that there's far more atrocities on the pro-IP side than the commie rat bastards they want you all to think us OSS people are. I could say generic "when you support..." joke, but its no joke. thats why so many of us here on
In SOVIET RUSSIA... erm...NSA AMERICA, the Internet logs onto YOU!
Large clients (e.g. banks) have the clout to ensure that once they select a hardware platform with a large provider like IBM, Dell, or Compaq, they will continue to get identical hardware on subsequent orders, even after the regular consumer can no longer order the components.
The same applies to the software they run. End of life to a large corporation only means that the general public can't get support for the product and is forced to upgrade; corps keep getting support for as long as they are paying enough.
Most corps I've worked for are running software that no one would even think of buying or installing anywhere else. It's all about maintaining compatability, and lock-stepped upgrades of entire farms of corporate systems. Even applying a software patch for the OS requires regression testing of third-party and internally-developed software that the OS vendor often does not have access to.
The last large client I worked for takes about three months to determine if an OS patch can be rolled out. Until then, you live with the problems caused by the OS bug, even if that means getting paged every morning to restart servers, or that users are going to have to put up with periodic dead sessions.
Absolutely nothing is more important to a large corp than data integrity. Not the sanity of the support staff, the profit margins of the vendors, or the "improvements" of a newer OS release. Nothing is allowed to change that might risk the data, and making changes without proper verification and authorization is a firing offense -- no matter whether they eventualy apply the update you forced or not.
I do not fail; I succeed at finding out what does not work.
Well.. the trouble is that OS tends to believe in release early, release often.
Now say that RH maintained an particular release. Over time they fix reported bugs, they update software when there are security problems, they update software for extra features and functionality, and so on..
Well isn't that just an upgrade?
However, a year is too short a time I agree. But 10 years for a server? That would require them to take every bug fix and back port it to probably 20-30 distro version. It would become totally unmanagable.
I feel for redhat and see the problem they are in. They have a very very large code base to maintain.