The Increasing Cost of Red Hat Linux?

An Anonymous Coward asks: "I work at a company with a large number of Linux servers in the data center. We're currently evaluating what distribution we want to use moving forward. Upgrading to Red Hat Enterprise from 7.2 would cost ~$350k just for the systems we already have deployed. Due to the change in Red Hat's release policy, we either have to move to Enterprise, or change distributions. Also, we don't have Oracle on any of these systems, but we will need it in the future. This leaves us with rather limited options. I'm interested hearing what other Slashdot readers are running, and planning?"

Debian!

Debian [debian.org] works well and the price is right! Wonderful install procedure too.

Re:Debian!

[To further save Slashdotters' bandwidth, another post about Gentoo zealotry was deleted from this space]

Re:Debian!

I think the whole point is support. With redhat, you're really paying for support - that's the whole point of paying.

With Debian, you don't get any support (IRC and google don't count when you've got to have a problem solved for your business in seconds. In these situations, "dial a tech-support number" == "support").

Re:Debian!

You can buy support for Debian.

Re:Debian!

You can buy support for Debian, yes. But when the packages will be updated... that's another question.

Now don't get me wrong, I like Debian, I use it on my personal servers. However I recall then whe last sendmail exploit came out it took a few days for the patch to be released. I tried to inquire when it would be coming out and was rudely told, "We don't comment on such things."

apt-get is a wonderful tool. But until patches are brought out in a more timely fashion I can't in good concience recommend it to any of my clients.

The other issue I've found, but I'll admit haven't put a lot of time into finding the solution for, is having a local mirror. When I build a server, if that server were to die, I want to be able to create the exact same version again of all packages. I have run into the situation with Debian of a package being upgraded and breaking things. Though as I said, this is something which is probably solved by now since I haven't looked into this issue in about 2 years.

Once these issues are resolved, go Debian go! I know we're facing the Redhat issue at the end of the year... upgrade all out RH7.3 machines to 9 by the end of the year... or risk not having security patches. I feel bad for the admin of the local 96 node beowulf. I'm glad I put RH9 on my new cluster....

There, my rant for the day. :)

Re:Debian!

One of the nice things regarding Debian is that basically all it is is a set of installed packages - no extra magic, basically. Creating a local mirror is as easy as creating a local APT source and storing all the packages there, and then instead of running apt-get update on the machines, run apt-get update on the mirror PC, which updates all of them. If the mirror PC works fine, then copy all the packages to the local APT source, and boom, you're fine. The details here are sketchy, yah, but it's an easy problem.

Regarding the security patches, I honestly don't know what problem you have with them: maybe Debian has really improved security support since then, but if you check Debian's page, you'll see that security.debian.org's response time is just as fast as any of the other major distros. There are several bugs for which Debian had a package that fixed the problem first (the SSH bug that required privsep comes to mind).

And honestly, I have NO idea what problem you had where a package broke something badly, unless you were running unstable. In my experience, Debian's packages are FAR less likely to break a system than some random less-0.4.3-mdk3-only-work-on-a-sunday.rpm. The few problems I've had were dumb problems that were immediately obvious (and in fact were stupid user errors, as I forced an upgrade of a package without forcing the upgrade of its neighbors).

I've never been happier since I converted my lab's PCs to all Debian. Yah, it's small, but I have to handle something like 7-10 PCs, and having them all in almost exactly the same state (which is far harder to do in Red Hat than in Debian) is SO nice.

I mean, the main reason Debian stable is farther behind than everyone else is because they take their time. When they mean stable, they really really mean stable - not just stable as in 'won't crash', but stable as in 'will do what it says it does'.

Re:Debian!

Yah, it took about 15 minutes of poking around apt-get's man page to find how to upgrade only the security packages by default (you CAN do this - apt-get update; apt-get upgrade;apt-get update with the options to select a different APT config for the first two, and create one that only has the security servers) - but it wasn't that hard, and now we've got automated security updating. Works for me!

I can not STAND people that say "Oh, well, it won't run on anything but Red Hat". Give me a break. The operating system is called Linux, not Red Hat (OK, maybe GNU/Linux). Linux defines the API and the application interfaces (ditto GNUification), and quite simply, everything that runs on Red Hat will run on Debian.

Period. Wackos who tell you "oh, maybe it's a problem with Debian" simply don't understand the way computers work. That's why I can't stand that Oracle won't support anything except Red Hat. That's silly. More than silly. They wrote a program, that works under Linux, not under Red Hat. If it's kernel version dependent, state the kernel versions it was tested under - or better yet, give the source tree! (wow!) If it's library dependent, give the library versions. If it's library dependent, static link the damned thing. There is nothing that runs under Red Hat that can't run under Debian.

You know what someone really needs to do? Write a bunch of scripts that let one distribution 'play' as another one, so you can just reboot and launch as a Red Hat clone, Debian clone, etc. (if you don't need a new kernel version, you don't need to reboot). It can't be that hard. That way when someone asks you what type of Linux you're using, you can say "What type would you like it to be, so I can then prove to you that you're being an arrogant prick and it really IS your problem?"

"distro-mode redhat". That'd be cool.

Re:Debian!

I guarantee I can make any RedHat application run unmodifed in any other Linux distro with a handful of caveats:

1. Both the host distro and the binary must be of the same architecture and must both be of the same binary format (e.g. two ELF-x86 distributions).
2. The binary may not use any system calls outside those required for the single UNIX spec. (This rules out things like ipchains/ipfilter/ipfw/ipfoo and various other kernel-version-dependent tools). This rule could probably be relaxed a lot before anything would break, but YMMV.
3. The kernel must be patched to fix any known bugs in SUS-compliant syscalls.

With the note that there may be other important directories needed in step 4, the basic procedure should go something like this:

Step 1: Install the distro.
Step 2: Install RedHat onto another machine and configure ssh and networking between the two machines.
Step 3: On the non-RedHat machine: mkdir /rhbox ; cd /rhbox
Step 4: ssh [e-mail] "tar -czf - /usr /lib /bin /sbin /etc /var" | tar -xzf -
Step 5: alias appname chroot /rhbox appname

Different operating system, my ass.

Re:Debian!

Each distro is a different OS in the same way that every installation of Windows 98 or 200 is a different OS -- i.e., the library (DLL) and package (service pack) versions are different from one system to another, and so the behavior of a single dynamically-linked application may vary across them.

However, while Windows application vendors are happy to support every version from 95 to XP, most commercial Linux applications are extremely specific about not only the glibc and kernel (more or less equivalent to the base Windows build) versions they support, they usually tend to refuse to support users under distros other than Red Hat. It's understandable from a revenue-based POV, since Linux as a whole probably consists (even for the most hardcore scientific or engineering app vendors) of less than 10% of their business.

Realistically, though, the effort and cost required to support at least the last few versions of all the major Linux distros (Red Hat, SuSE, Mandrake, Debian, Slackware, etc.) is probably less than the support for Windows 95, 98, ME, NT 4, 2000, and XP. It's not a tecnical issue so much as an economic one, but it does negatively affect the natural competition that exists in the Linux distribution market, since any new vendor has to either work towards 100% compatibility with recent versions of Red Hat, (and therefore use RPM, standard SysV init scripts, etc.) or accept an extremely marginal, source-package-only application support model.

Re:And also,

Just a little info for you:

Postgres is one of the oldest and best supported dbms around. It doesn't have a couple of features Oracle has, but it is a highly superior product compared to most others in it's class. That's in a business setting anyways. MySQL is easy to use, relatively fast at simple operations. DB2 is nice. Good for big computers with lots of ram. That's my observation, there probably is no rhyme or reason to it. I have lots of ram, and I like DB2. It's just fun! MSSQL sucks rocks. I developed a lot of websites using MSSQL as a backend and just couldn't make it feel right. Oracle is great for big iron.
They say it's good on x86. I say Ellison is full of shit.

Here's my personal dbms preference list:
Business
DB2
Oracle
Postgres
MSSQL
MySQL
Hobby/Pleasure
MySQL
Postgres
DB2
MSSQL
Oracle(unless you are masochistic)

Re:Debian!

Hello, enterprise Debian user here. :-)

We manage consistant, reproducable server installs by NOT using the Debian installer. We install it once, and then make tarballs of the install, which we untar to install the server. We have it all scripted, so we can boot a machine off of a CD, run a script, answer 5 questions and have a server ready to go in about 10 minutes, and have the applications working (also script-installed) in another 10 minutes. We maintain our own OS release numbers (versions of our base tarball), and our own .deb packages of our applications.

For mirroring: We use debmirror. (It's a Debian package, of course.)

When upgrades fail: Go into aptitude, find the package that the upgrade broke, scroll down to the bottom, and highlight the old version, hit + to install it. It'll gracefully downgrade for your convenience. It's a hell of a lot easier than downgrading on, say, Solaris.

--Keepiru
--slashsuckATvegaDOTfurDOTcom

Re:Debian!

Hey, have you ever tried System Installation Suite [sf.net]? IBM really pushes it for clusters and large installations.

Re:Debian!

Well, the $350k/year he is talkin about can hire quite some 24/7 on-site support. For Debian too. Just put some good people on the payroll.

Re:Debian!

$350k/year won't really get you very far at all for a well-staffed IT department. You're going to end up paying out the nose for any systems administrator who specifically addresses critical problems, as opposed to the more mundane IT staff who may deploy patches, ensure systems are running properly and not experiencing memory leaks, random errors, etc. Assume that you're paying $70,000 for a Linux sysadmin who's experienced enough to handle all the problems that none of the other guys can manage on their own (which is the entire point of the pay support in the first place). You can hire five of these people. Splitting them into 24/7 support (3 shifts, plus weekends), you essentially have one person at any given time managing 500 servers ($350,000 divided by $700, the cost of a Red Hat Enterprise ES license).

Outsourcing is really a much better option than hiring these people.

Re:Debian!

Actually, you could only hire about 2 $70K sysadmins for that money. An employee usually costs close to 2 times his/her salary.

Re:Debian!

Faster? It is often easier/quicker to just use Google to search for the answer.

1) You don't need wait while being tossed between technicians.

2) You don't have to wait for a callback.

3) You already know the details of your problem. You know what you tried. You don't need to try and communicate these to person on the other end.

What types of problems have you come across that have been handled better by tech support people?

Some have mentioned needing to wait for an updated package to be released by your distro. Just grabbed the latest source/patch and compile it yourself. Can't get too much quicker than that.;)

Re:Debian!

With redhat, you're really paying for support - that's the whole point of paying.

Actually, no. That *used* to be the case, but if that was the only thing, it wouldn't be that expensive.

What you're paying for is certification. Wanna run Oracle? Well, if you want to get Oracle's "unbreakable" support, you have to run it on a certified OS. Getting that certification costs $$$, and lots of 'em.

It's the same with a number of other production applications - if you want support from the vendor, you have to run an OS that they support. If you've got your own home-grown kernel and you start having problems, how are they to know that some crap you put in there isn't hosing things up? And it's certainly not feasible for them to support ever kernel-(user) release out there - so they pick the biggest fish (RedHat).

For your servers that don't run production systems, just use the regular Red Hat and buy support (if not from Red Hat, then somewhere else - the beauty of open source). Or run Debian/SuSe/Mandrake/whatever. Doesn't really matter, so long as your *production* machines are certified.

Re:Debian!

For your Oracle instances, I HIGHLY recommend their Advanced Server. I've seen it in action and it is impressive. I would consider it to be worth the cost.

For everything else, CHANGE distributions. SUSE, Debian, Mandrake, ASP, Rawhide, pick one. Or ditch linux all together and use FreeBSD. If you guys are used to Solaris, FreeBSD will be a very simple transition. The other BSD's are good too, but Free is closer to Solaris, IMHO. I've found that my experience with Solaris has translated to it quite nicely. In addition, the documentation is fabulous.

HTH,

Queen B
Jimi

Re:Debian!

If you guys are used to Solaris[sparc?], FreeBSD will be a very simple transition.

Orr.. Geee.. run Solaris (x86)

Re:Debian!

Ok, who rated parent is "funny"? Debian is the only real option IMO for using Linux in the enterprise. This is because of it's testing, stablity and the sheer number of platforms it runs on. Plus, you never have to worry about "purchasing" newer versions. Red Hat is often released very quickly when the software which is is made up of, is often not thoroughly tested. Yes it's the bleeding edge, exactly, it can indeed leave you bleeding.

Re:Debian!

That got moderated funny?

I work in a mid-size enterprise, and we're using Debian with great success, for most of our unix needs.

The right distribution is highly dependant on what your specific needs are, though. Here are some things to consider:

1) Debian doesn't provide direct commercial support. Tech support is available from third parties, but by the time you get a support contract that will equal the quality of support you get from Red Hat, it'll probably cost as much as Red Hat (or Solaris, or Windows, or anything for that matter). Saving money is a priority for us, so we simply self-support. We maintain our own baseline install, and take care of certifying our own apps against Debian as needed. It's worked very well for us, but if you need CYA, Debian may not be a good choice.

2) Debian's release cycle is too fast. Yes, I know you hear it from people on /. all the time that Debian is so out of date and how terrible it is that they only get one release out the door per year, if that. However, in the enterprise, upgrading everything once per year is painful. In an enterprise, you can't just change your apt sources and upgrade in place. We get a window - say, one hour - to perform an upgrade. That means that I have to have a machine built and ready to go, other than syncing the data across during my window, and get it back up, and have a way to fail back (still in that window!) if something goes wrong. This means I have to rebuild every Debian box in the company once a year. (Because we standardize our base install, and have scripted all of our application installs, building new machines is extremely fast for us, to the tune of 20 minutes or so, but it's still a lot of work to re-QA our applications once a year.) For this reason, we only use Debian where version upgrades go pretty easily - BIND, Apache, Postfix, FTP servers, etc, where the Debian-supplied versions consistantly work right out of the box. We use Red Hat Enterprise for complicated commercial software (Oracle), where the long support cycle means we only have to go through the fun of upgrades once every several years.

3) Oracle doesn't support Debian. Sure, it installs, sure, it runs fine, but that doesn't mean it's officially supported. This means that when you call their tech support, they will laugh at you the moment they think it's an OS problem. And, despite being head over heels in love with Debian, I think Oracle made the right choice - Enterprise customers who are going to install an application that needs to be in use for several years need to evaluate several other factors than just how easy an OS is to install and even how good the quality of the software is on the OS. I firmly believe that Debian builds a better mouse trap, but for Oracle, I want an OS that has official commercial support and a long life cycle - Red hat Enterprise.

Despite all this, don't think I'm trying to say that Debian is a bad choice. As I said at the beginning, we use it extensively for many functions throughout our enterprise, with tremendous success. Regardless, you need to evaluate your needs against ANY distribution, to see if it's a good match.

--Keepiru
--slashsuckATvegaDOTfurDOTcom

I'm Running Windows XP

And I'm planning to go home and play America's Army.

I may need to reboot 3-6 hours from now, but I've never had to learn how to edit a configuration file.

(Disclaimer: That's not really true, but you get the point.)

Re:I'm Running Windows XP

Yeah Microsoft sure is great for playing games isn't it. Yep, it sure is great for playing games.

Re:I'm Running Windows XP

I've never had to learn how to edit a configuration file

I agree. Having all the configuration information, for all applications, bundled up with lots of other stuff in that registry makes editing so much simpler snd safer.

Enterprise != Free

Usually, enterprises aren't interested in free or next to it. They want stable and supported for a stable price.

Re:Enterprise != Free

Ummm, IBM supports [ibm.com] Linux on its top of the line zSeries [ibm.com] mainframes which pretty much define enterprise class.

Re:Enterprise != Free (but you can negotiate)

At $350K to upgrade, you are talking a serious number of systems. I'm not saying you can talk them into giving it to you for free but whoever is doing the purchasing should be able to negotiate something better than full retail. You are mainly buying support so things to point out include multiple identical systems, internal support for end-user systems, etc. that mean they won't have to answer too many really dumb questions.

Favorite really dumb support question: do I have to plug it into the electricity?

Suse ?

How much more would Suse cost? I have worked at facilities before that switch from windows to Suse recently and they said it was a lot less expensive in the long run.

Re:Suse ?

We have been using RH 7.3 for some time now in our cluster configuration.
Now we wanted to upgrade for various reasons and we really dont want to have to do the upgrade manually each time so we looked at RH. Damn its expencive. It's way to expencive when you have to pay a license pr. server, so we switched to SuSE Enterprise server (Oracle Certisfied)
Now we do automatic upgrade/patch of all servers using only one License. Also everything seems to be running much more stable. We have a FC connection to an external diskarray RH crashed several times using this array, also the webserver made some strange hicups from time to time (Also with RH9), but after changing to SuSE we're moved away from these problems.
Also.. when talking about support. RH is difficult since they only reply to web support, slow and it's very very hard to describe the problems for them. SuSE on the other hand replies to mails, are fast and very helpful and service minded.

Re:Suse ?

I first want to say that I think Redhat is insane with their new pricing. I understand that they want to make money, but they need to make some serious changes to their current model.

Next, what you said about email support only isn't true for the Enterprise version. You get to call them, and their support is good.

I am in the exact same boat as the guy who posted this, and considered SuSE and RedHat. Here is how it broke down for me. I also need to run Oracle...

RedHat
$350 / server without phone support or upgrade protection, but you get up2date for a year, and some basic (email) install support. In my opinion it makes no sense to buy this version, given that 3.x of their product will be out this year, and that version will offer things like LVM install built in (and a lot of other things), and you would be forced to buy that version for $350.

$800 / server with "Normal business hours support" and upgrade protection for one year. This version is limited to 2 processors and 4 GIG of memory.
RedHat had more expensive options but these two covered my company needs.

SuSE
$700 / server per processor. Support included. Also Opteron support.

With both products the cost seems to be for support and you will need to pay them EVERY year some amount of money. I would put that amount down, but my belief is that it will change given market demand. Also it must be noted that neither one allows you to load a copy of it on a "test" or "development" box! You must plunk down the $350-$800 again!

What I wish RedHat would do is allow you to download the Enterprise edition and install it on as many machines as you want for a nominal price. Say like $300 / processor. (NOTE don't limit processors/RAM ect) but pay for Up2date and then offer a support packs. Specifically they need to offer like 10 calls to them for $5,000. Novell and Microsoft currently offer agreements like this and it works well. They could then offer a pack of 50 for a discount and so on...

When I approched RedHat about this they seem to believe that their competition is Sun and even with this pricing they are still cheaper than Sun, so it makes sense to them. I don't agree with this! They may take away 10% of Suns business, but they WERE taking away more than 1% of Microsofts business, and the fact is that 1% of Microsoft's share is a heck of a lot more than 10% of Suns.

Now one last point in favor of RedHat. Oracle DB Standard Edition charges $15k/cpu + support + maintanence. That comes to around $22k/processor then you have to pay support + maintanence EVERY year. That comes to around $5k EVERY year to Oracle. If your company can handle that, then throwing RedHat a bone every year shouldn't be too bad. It just seems extra bad because it use to be near free.

Also, DB2, Websphere and other IBM apps have the same requirements as Oracle in regards to Linux distros, so it appears that the "big business" has kinda dictated what RedHat and SuSE will do.

the real question is...

does that cost include count the SCO license?


oh come on, thats not flamebait!

We run red hat

We made the upgrade. Its a godd choice. You know what you get, you get oracle/ibm/big gun stuff.

AND

you support open source / free software.

Re:We run red hat

you support open source / free software.

Yea. Try explaining to the brass that it's going to cost you $350,000 and it's free software.

Changing distros?

Have you evaluated the cost of moving to the supported versions of SuSE, etc? What's the cost there? How does it compare to Red Hat?

Also, if you find you don't need support, then why use the "enterprise" editions at all?

Finally, what'd be the total cost of moving to Windows? Probably a lot more than $350k, I'd wager. It sucks, but it's probably just time to pay the piper, or deal with supporting yourself... that's just how the market is. RH have to make a profit somehow.

What do you need it to do?

I think the first thing that should be asked is, what do you need to do with it? Distros have a strengths and weaknesses. If you just ask, what distro, you end up with a giant flame war over which distro is better. Also, Have you considered possibly using a version of *BSD?

Re:Cost explanation

Buy one support licence and use it on all your machines, its GPL you don't need hundreds of copies of RH Advanced Server.

But this puts you in violation of your support license, resulting in its termination and therefore not being supported if they catch you.

However, this will result in the same level of support as if you still had a support license.

Re:My reccomendation?

For $300k/year I will support it.

My recommendation...

Never take recommendations from anyone who spells the word with two c's and just one m.

Laugh stupid, it's a joke.

What are your goals?

If you're looking for support (which is what I'm assuming your reason for going with Enterprise server is), then either pay for Advance Server or go with a different cheaper distribution and put the money you saved into someone that can search Google and find out how to make "RH only" stuff work on Debian or something.

We run oracle (both 8 and 9) on Debian, as well as most of our internet infrastructure (with the exception of proprietary programs that are stuck on Win2K for the time being). Most of the vendors of Linux based apps that we have worked with are willing to provide support even with Debian being the distro we chose (and then the ones that have complained, I've just called for another technician that was more distro-agnostic and gotten right through).

Recommendation

Debian. Or Slackware. Just be sure to have a copy of either Knoppix or Slackware Live CD handy. Write down what the CD auto-detected. Select appropriate odules when installing proper distro.

And MySQL for the database.

Redhat too expensive?

Switch to BSD! I hear FreeBSD is nice. Also in the enterprise. And a license that does not make $neckties nervous.

Re:Do you really need RH Enterprise Server

You are high, or highly misinformed.
RedHat continues to develop a public release
the beta for the next release (redhat 10) is available right now for public consumption. The simply aren't providing shrink wrap at compusa anymore, I think what the original story is referring to is the fact that each public release is only supported for 1 year, which is unacceptable in a corporate environment, to have to upgrade the OS of production boxes every year is not acceptable, and therefore requires that people move to ES, because it has 5 years of support.

You dont have to buy it

You can just make your own build of redhat. Every piece of the OS is available as source rpms from redhat themselves, for every linux OS they sell.

Get em, compile em and install em. Of course, the nice gui installer is not free, nor is the support. But updates and the OS itself is free and will always be free. Its GPL'd. What you pay for is support and peace of mind. Thats typically what data centers prefer these days. I know that the managers see only free as in beer, so they look like heroes for saving on the budget, but what really counts is uptime and reliability. TCO stuff. So it costs 350K... How much would Windows cost you, and how much functionality would get from it? How about the equivilant PA-RISC machines or big AIX boxes? E15k's?

It turns out to be quite a deal! The support you get is worth it, and compare the price of that to a support contract with Sun!

Re:You dont have to buy it

> Of course, the nice gui installer is not free

Sure it is. It's called Anaconda, and everything you need to make your own customized version is included in the anaconda package in Red Hat, licensed under the GPL.

You're probably thinking of SuSE's installer.

Is there anything to stop you copying RHEL?

I've been trying to figure this out, and it would seem that there is nothing in the license that stops you being able to legally give me a copy of (say) RHES, and for me to run that copy, with no access to up2date and no support contract. (Like a lot of the other posts say, I am the support - my only concern is having a platform that commercial software supports!)

The license [redhat.com] seems to refer to the services that come bundled with the software, not the software itself. I believe that the JVM cannot be copied from the standard distribution but removing is trivial.

Interesting notes: [linuxmafia.com] to summarise, it's probably perfectly legal for you to copy me RHEL ES, however you would probably also have to provide me all the updates if I wanted them (which may violate your license to receive them). The big dollars is with regard to the updates, and I believe they are made publically available by SRPM - and even then, its probably also technically allowable for you to mirror all the update RPMs somewhere.

I installed Lotus Domino recently on a Debian server because I didn't trust the machine with a consumer Red Hat and it wasn't cost effective enough to get RHEL. I'd be very interested to hear if you can or can't just copy/mirror RHEL.

What is the software worth to you?

You need to look at what you are paying for, and what you need. With Redhat you're paying for a package (eg, physical box of stuff), some of their packaging expertise, a small amount of their own custom goo, and presumably support. You're also indirectly funding GNU and Linux development. If that's not worth $350k, there are a number of options out there.

I personally use FreeBSD. No, I'm not suggesting you switch, but since I use it I'll detail it as another point of view. I download the software, for free, and pay no licenses. I also don't get a pretty box, support, and I've done nothing to fund development. The pretty box is available, for a fee. Support is available from a number of companies, for a fee. You can fund development as much or as little as you like with donations.

Without telling us what you need, we're not going to be able to make a recomendation. Maybe you use some Red Hat "feature" a lot that's worth $350k/yar, maybe you don't. What I can tell you is there are more expensive (price Microsoft!), and less expensive (eg, FreeBSD) options. There are also many, many, many options in the middle.

I suggest...

If you are sick of RedHat's extortive licensing fees that you instead switch to Windows XP... :)

$300,000 worth of support contracts.

That's esentialy what you are paying for if you go with the RedHat enterprise. The assesment you need to make is;

1. Do you need that level of support.

2. Is there a cheaper way to achive the level of support you do need.

3. Dose 1 or 2 requiere switching vendors.

For the cluless. It has nothing to do with the software itself. I.e. You can download RedHat and install it on as many PCs as you like virtualy free.

PS: Support for large numbers of critical solaris and/or Windows servers costs just as much or more. I should know since I work for a company that makes most of it's money off this sort of thing.

Cha-ching!

At my company we had been using IBM's advanced AIX support for our 60+ RS/6000 and pSeries server environment. The cost? Roughly $10K per month. We were typically calling them once or twice a month and there would be two or three instances during the year when we would ask questions to which they had no answer. Needless to say, we are no longer using them. I mean, would you spend $120K/year for someone to sit on your staff doing nothing aside from answering two questions a month, even if he couldn't answer the questions? Be real.

I'd consider this when getting the level of support you have priced from RH. Think about it: will there be many questions for which you are willing to pay $1000 a pop? Are there many questions to which you couldn't find an answer by Googling? Or is it more of a CYA action in case your team fouls up? I'd hate to think you'd be wasting over a quarter million dollars for an inept admin.

I still like RedHat... so here's what I do.

The strategy I'm taking is to use the RedHat high end products for running commercial applications (like Oracle). For everything else, standard RedHat is fine. The developers like that RedHat keeps pace with the new stuff they're working on.

The advent of http://www.fedora.us bodes well for the future. I expect to see more 3rd party support for the RedHat standard package. That's the nice thing about RedHat finally opening up their devel process.

At worst, you could just take the standard distro that RedHat bases their advanced products on and use the security patches from the advanced on the standard distro. For example, install RedHat 7.2 and install any patches from the currently support advanced product. The only thing is that you'll have to rpmbuild --rebuild the src.rpm's as they are released.

I really like RedHat's way of doing things. I like their python based configuration programs. I like /etc/sysconfig, /etc/profile.d, etc... very modular.

And who'd 've thunk... RedHat is basicallly IPv6 ready out of the box. I didn't notice that until recently. Very easy to setup 6to4, radvd, etc. Even Mozilla is compiled with --enable-ipv6. Thanks RedHat!

Re:I still like RedHat... so here's what I do.

I agree, I think Red Hat is really leaving small/medium businesses out in the cold with their new policies.

One example, I set up a server/firewall for a small business as a contract job on the side. It runs all their stuff, web, email, etc. They really don't have a need for more than one server.

Last year when I set it up, I told them it would be good for at least three years. I don't like it that Red Hat made a liar out of me.

At my real job, we have more than a dozen Red Hat servers, and I'm stuck in the same dilemma. We can't afford to destabilize servers by putting essentially beta software on them, that the 12 month EOL requires. I used to let RH releases age about 3 or 4 months to get the major bugs out before I even considered using them in production.

So now I'm faced with either spending tons more time patching the servers by hand (even though we pay for RHN), or spending tons more money, buying RHEL which would cost at least 10 times more than the ~$1000 a year we spend on RHN now.

I really hope something gives on this situation. Red Hat better think twice before they alienate small to medium businesses. I don't need support. I do need updates. I do need more than 12 months before EOL.

Thats the point, idiot

Red Hat are in business to make money - they do this by providing paid-for distributions with full support, custom-tweaked kernels and applications, and provide a validated platform on which to run commercial apps like Oracle.

Want to put that together yourself? Go for it, nobody at Redhat is stopping you. All the stuff they integrate in their product is free, just go do it yourself.

But don't complain because you can't do it yourself and don't see why you should pay Redhat to provide you with a quality product.

Its not like you don't have a choice of vendors, or that your apps only run on a single vendor's platform.

Linux is never free - you either pay for it with money, or you pay for it with your commitment to the GPL and/or the time you invest into making it work for you.

We need people like you in the Linux community i.e. 'waah waah linux is too expensive, even when i can download it for free' like we need a frickin hole in the head.

What do you want?

What do you want? Enterprise-level support without paying for it? Do you think that the support contracts offered by HP, IBM, Sun, or Microsoft will have more value for less money?

This is the Free Software movement, not the Free Support movement. You can still download the software for free, and pay some kids $20 an hour to support it if that's what you want. Quit complaining that the world doesn't give you everything you want for free.

The value of Red Hat for an enterprise is not that the software is free of charge. The value of Red Hat is that the source is free from restrictions. Other than that, they're just like any other enterprise Unix vendor.

Re:What do you want?

Hell yeah brother! Preach on! I really don't get what all the bitching is about. $800 for 5 years of support? I support 13 app servers that we pay $16K/yr for support on. That doesn't include the OS or hardware. So if Red Hat is basically letting you have the code for free but only answering the phone or providing patches if you cough up a little coin how can you complain? One dude said he only called IBM twice a month. If you called Red Hat twice a month and say each call lasted about an hour you would take up three working days of that guys time a year. If the tech support guy makes $40K/yr and you add in 30% for benefits and then some for cube rental they aren't making a fortune on this deal. Plus, you would have consider how much a developer might cost to write a few patches. I think the problem here is that people are looking at it from the perspective of the home/small business user and expecting the same sort of support one would get from an enterprise software vender. From my experience that kind of support can cost tens of thousands per quarter. Someone running SAP, peoplesoft, or Oracle shouldn't blink an eye at $800/box. If you think this is too much how do you expect Red Hat to make a profit? As a shareholder I'd really like to know.

Is homogeniety a requirement?

You want to run Oracle with Red Hat, period. (Well, you want Red Hat unless you want to pony up 250k for a sunfire box, or go back to 2003. Saving money is the objective, right?)

However, you can safely run other servers on GPL Red Hat, or Debian, or SuSe, or....

We moved to BSD for most of our Unix needs.

an explination to this seemes merited

Upgrading to Red Hat Enterprise from 7.2 would cost ~$350k just for the systems we already have deployed.

Someone please explain this claim. I have no experience with buying anything from Red Hat, but I was certainly under the understanding that the software was freely copyable. Further, if you bought one copy you should be able to install it on as many systems as you wanted. Sure, support is an issue. And if you want Red Hat to give lots of support for a lot of systems you should expect to pay for it. But couldn't AC and his company hire more people and support the systems themselves with that $350k? Don't they need support staff anyway to work with Red Hat? They would have to have support staff if they moved to Debain or other distros, so is there really a reason to move rather than stay with Red Hat and support yourself? Is there something about using Red Hat that I'm unaware of? Where is this $350k cost coming from?

Re:an explination to this seemes merited

Please read the redhat AS license. This is where the costs come from.

If they would simply put the offensive language in the SUPPORT CONTRACT where it belongs and not in the OS license where even the leaders of Open Source and linux find it offensive.

Basically, they have added things that make it look like a microsoft product license..

Please read it, it's online for a free read. It will upset and enrage you.

and It's the reason I have migrated my company away from redhat on it's servers to Mandrake.

I'm all for paying for support, I have subscribed to redhat support in the past, hell I owned stock!

But redhat is pissing on those of us that made them what they are today with their insulting license.... and that is something that doesnt sit well with me.

Offering support is one thing. Forcing me to buy it is another.

Unless someone else can tell me how to get my hands on Redhat AS without paying for the support, it's not a viable option for any enterprise that has skilled staff.

Re:an explination to this seemes merited

What the hell are you smoking? Why don't you actually READ the license?

Red Hat Linux Advanced Server 2.1

How exactly does the license prohibit free copying or redistribution of their product? As far as I understand, if you don't want support, you can just buy one copy and use it on all your computers. Or use somebody else's copy on your computers. You can even redistribute it if you remove redhat's trademarks. This is way, way, way better than what SuSE offers.

BTW, Mandrake uses a practically identical license for their server products.

Redhat's support sucks

By all means, get Redhat support if you're just trying to make your company feel good about spending money on something. But their support is terrible. By terrible I mean completely worthless at solving any sort of problem easy or complex, big or small.

They aren't much worse than anyone else's support so far as I have experienced. But still somehow I was shocked at just how completely worthless they are.

Moving forward?

We're currently evaluating what distribution we want to use moving forward.

Well, you could install just about any distribution on a laptop and hook it up to one of those Evolution Robotics laptop robots [evolution.com]. Those go forward (and backward, and sideways) quite easily. Oh, you meant in the future? Well, why didn't you say so?

Pointy-Haired Boss: "We need to do this on a going-forward basis!"
Dilbert: "Thanks for ruling out time travel. You're usually not that helpful."

Cheers,
IT

Independent Linux providers may be of service

My company has hired a small independent Linux technical support provider to help with this.. They have a service where they create patches and updates for RH 6.2-8.0 when new security vulnerabilities are out. They test them, package them as RPM, and distribute. So when 12/31/2003 comes around, you don't have to upgrade to 9.0 if it isn't feasible.

Its kind of expensive and may not work for everyone, but its worth a look:

http://www.pantek.com/linux.php?subsect=rhupdates [pantek.com]

In this economy when the "big" Linux players are worried about the "big" issues, I prefer working with a smaller company like these guys because they work harder to make their mark.

We left RedHat...

...for Debian.

I'm not saying it's the answer to your problem, I don't know, you'll have to decide.

Now, before we move on I'm going to tell you how Debian sucks. This is not to say that other distributions do not suck, or that Debian sucks more or less than the others - this is just something that you might run in to and should be aware of.

Debian sucks because:

• It's a pain to install (no software RAID support, default kernel is 2.2, yadda yadda)
  
• All packages are *old* - it's hopeless for a desktop
  
• Fewer commercial packages available (suckage when you need them)
  

Yet, we chose Debian because it rocks (and RH sucks) in these areas:

• Updates. Usually there are no updates to the stable distribution except for security fixes. This is *very* good when you actually have to maintain your systems.
  
• Updates. "apt-get update; apt-get upgrade", and voila you have a list of security updates available - and you're about to install them. No subscriptions, no fees, no wondering where to get them from. It just works.
  
• Simpler package dependencies - it is actually possible to configure a web server without installing GNOME (ok, this particular setup is *probably* still possible in RedHat) - in general you will find that for dedicated servers, you end up with a 100-200 Meg system where the RH system it is replacing was well over a gig.
  
• Clear roadmap. Who knows where RedHat is going? Debian is going nowhere, or at least they are moving very slowly - this is actually a very *good* thing in this respect.
  

For a server you put in a data center and don't want to touch again unless absolutely necessary, I think Debian is great. It is extremely easy to stay up to date with security, and that is pretty much all there is to it. I still have nightmares from the days where I was mirroring entire RedHat distribution trees (or at least their massive update directories) in order to keep those systems up.

But really - in the end - it is not a few hundred bucks per server that should make the difference. It is my impression that if you pay for your RedHat, you can have a nice update service as well.

You'll be shelling out thousands of dollars per server for the hardware, an order of magnitude more (over the years) for support (eg. your time), so a RedHat subscription fee really shouldn't stop you from going RH.

On the other hand, if some of the above made you think - I can promise you that Debian certainly is a viable alternative at least for the machines I've dealt with so far.

RedHat...

I personally believe that Slackware is the best server implantation that you can go with. It's the most Unix distro IMHO. Also you deal with source files, not RPM's, personally I greatly dislike the use of binaries. The best thing is that Slackware is Free, and it's stable. I know a lot of different people whom have had problems with Redhat, switched them over to Slackware, which has a slightly higer learning curve, but, they were happy with Slack.

Debian is an obvious choice

Debian stable or testing are reasonably up-to-date, regularly updated by volunteers all around the world, and entirely free. Not only is updating to new versions a lot easier than with most other solutions, Debian packages also come with nice configuration scripts which make your work a lot easier. For servers, I really see no good reason to use a large, commercial distribution like Red Hat. For clients, Debian is a bit too outdated (unless you install lots of backports or use unstable, the latter not being an option for companies).

Support, you say? Debian has a nice directory of qualified Debian consultants [debian.org], and in general, it makes sense to have a few Linux experts inhouse to deal with emergencies.

How much support do you need?

How many hours of Redhat support did you use last year? Divide the number of hours into the support contract cost. If the hourly rate is over $100, (and I'm betting it will be way over) consider getting on-demand support from independent consultants, instead of using a pre-paid contract. Some consultants will even let you buy reasonable (e.g., 10 hour) blocks of support time, which you can use in small (5-15 minute) increments. You have MANY support options. Explore them to see which will save you the most.

Use any savings for training. As your in-house expertise increases, your support costs will decrease. The nice thing about Linux is you only have to pay for the support you need. Too many companies forget that.

Learn to use Google effectively too; 99.9% of all Linux questions I get in a year have already been answered, and are just a quick inquiry away.

SuSe.

SuSe is cheaper than AS, how much cheaper i do not know. but unlike most distro's they offer an "Enterprise Edition." [suse.com]

They also offer priced to fit support, and now have the backing of IBM and Sun, and they support oracle.

and this is coming from a Gentoo zealot.

Call sales

Have you considered calling / writing to Red Hat's sales section with your concerns? You may be able to negotiate a more acceptable price. Especially where there's such a significant sum involved.

Why exactly do you need RH AS or the equivalent?

The new "low cost" server option at my company is RHAS on a dual Xeon box connected to a huge EMC disk for network storage. F*$%ing overkill, bigtime. And they wonder why we can't do anything cheaply. This is the small, minimum production grade server standard embraced. JHMFC.

In my opinion (not so humbly, though), the only thing you're getting from big, expensive RH is the guarantee that Oracle will support whatever f-ed up configuration you come up with. It's still GNU/Linux at heart (there, RMS, ya happy now?) Sure, RH promises not to change it as often, but honestly I just upgraded an old RH server running 6.2. It's been running and stable for something like four years. It worked, so aside from patching and security, I left it the hell alone. This is something that large companies can't understand. Once it works, don't upgrade every damn chance you get - keep the old solid configuration running until you have the time and the need to do an upgrade.

Personally, since I believe that having three truly hard-core linux geeks that know their shit onsite is better than any professional support line you could ever call, I'd go with standard RH and order me some geeks instead. For $350k, you should be able to get a very nice set of them, and they'll be right there to save your ass if anything goes wrong.

This is why I have no future management prospects. I just can't think that way - I worked in small shops too long to think that throwing money at stuff fixes anything. We found ways to keep stuff running on a mix-and-match room full of old hardware - no support contracts, no officially supported configurations, just guys (and one lady) that knew what the hell they were doing. Once I moved into the big corporate world, I had to give myself a lobotomy to even understand their mindset towards problem-solving.

Red Hat Sales is the worst I've ever dealt with!!!

I was personally involved in porting our company's software to Linux. I chose to support Red Hat, thinking that their big name would mean that they were somehow better as an organization.

I WAS TOTALLY WRONG!

I recently tried phoning Red Hat Sales to try and buy support, and it has been more than 1 week, and I have been unable to get them to respond! My first 3 attempts to contact Sales were ignored, and finally I got someone on the phone. They directed me to someone else, and after an initial e-mail, they have yet to contact me after I sent them 2 follow-up e-mails. It is absolutely ridiculous.

You would in this day-and-age that Red Hat would be salivating over someone who is willing to pay them money for support, but they seem competely disinterested in helping me give them money. I have already complained to my superiors that we should consider supporting a different flavor of Linux, because if this is how responsive Red Hat's Sales unit is, imagine how unresponsive their Support unit it.

Is this a sales quote or just his calculation?

I haven't seen it mentioned anywhere whether this is a sales quote or a quick calculator job? Sales people are flexable, especially when it comes to big sales. If your calculating 350K there is probably a Red Hat representative out there who could make you very happy (and would be very happy). Of course if your company needs 350K worth of support, then you'll need to be more creative.

But this sound like someone who is quick with the calculator and just as quick to react.

Dissatisfied with Red Hat? Try Gentoo.

I work for a medium sized (137 employees) company that processes customer data for many retail outlets, as well as a multi-national bank. We were one of the first companies to drop our entire line of Windows servers (workstations unchanged) for a Red Hat Linux solution in the summer of 2000. Porting our internal applications was a real pain, but the significantly increased uptime and greater ease of administration made up for all initial shortcomings.

Fast forward to end of 2002, and we had become disgusted with Red Hat's road map for its' Advanced Server license. It seemed as though we had lost all of the benefits of the GPL.

There was no way we were going back to M$, but there was a movement from higher up top to change distributions. To make a long story short, we passed on SuSe and chose the often corporately overlooked Gentoo.

The benefits of this move are stunning. We have been able to hire 16 additional employees to handle our own fork of Portage, and 22 additional employees to provide support. Not only to we do a "ghost compile" for each box (many different Pentium and Athlon systems), we also take a minimalist approach. The combination of those two choices have enabled us to increase performance per box to something like 26% faster on average.

With the obvious help of the Gentoo open source community, we have created a low cost, self-sustained IT department that can function well into the next decade. Thanks Gentoo!

Do you really need support?

I get more *support* from Google than any distributor can provide. In other section of my University, they call support for the stupidiest things like compiling problems, tape problems, its ridculous!! Why get support when it will be Apu on the other line?

Here's what you do:

1. Hire 1-5 high school Linux geeks part-time.
2. Pay them 15-20k a year. They will rejoice! Sweeten the deal with an unlimited supply of Twinkies, Mountain Dew and Hot Pockets.
3. Sit back.
4. In your next conference with the big cheese, tell him how smart you are for solving the company's IT problems.

FreeBSD

You can buy great support from BSDI. Of course the isp market is what they are more catered to supporting but the price is free and many consider it more stable then most linux distro's. FreeBSD developers are as picky as slackware and debian ones when making sure their distro is stable. Suse, mandrake, and even redhat or more cutting edge and the packages are not as well integrated and done compared to FreeBSD or Debian.

Yahoo and Microsoft's own hotmail run on FreeBSD. Also the apache team uses FreeBSD as their os choice on their servers. FreeBSD handles large amounts of i/o and tcp/ip traffic and that is its strength's. SMP support and threading are its weaknesses.

Just my two cents.

Also you can run Oracle if you install the Linux abi package on FreeBSD.

depends upon budget needs

It depnds upon your budget needs..

You have already mention the need to use Oracle which is $5k per cpu..

If you switch distros you want to make sure you are not placed in the saem decision 6 months or 2 years down the road again..

Thus I recommend looking into using debian instead of SuSE..

As you know SuSE is partnering with vendors such as Sun and thus wil be in the same bussiness model as Red Hat in wanting to charge for enterprise versions either through support or etc..

Do not forget to factor in training costs of employees of going from distro to the next...

Also do the idiot checks, take a demo/dev version of Oracle and do test installs on the distro your considering..

Is Redhat is missing their target market??

I work for a software company whose product is bundled with Redhat Linux and HP Proliant servers. The recent Redhat changes are bad news for our product. For the past few years, we've sold the HP/Compaq servers with appropriate versions of Redhat (7.x, 8) and our proprietary software on top. Redhat 7.3 and 8.0 have proven to be the best match for our software/hardware solution. The hardcore Compaq/HP server hardware support (for ML370's and ML570's) is there. The OS is stable. We use up2date to keep on top of security patches (openssh, etc.). It was nice because we could give the customer a real Redhat box with media and manuals (not that they used it... but it's nice to have the packaging).

Now, I have 70 Linux servers around the country, and a steady stream of new customers. I've been installing Redhat 8.0 on new deployments because 9.0 doesn't work well with our application. So, we've everything from 7.0 through 8.0 in the field. Over the past few months, Redhat dropped up2date support and patches for Redhat 7 and 7.1. I feel guilty installing 8.0 on new boxes because I know support for it will be dropped at the end of the year.

I don't wish to buy into Redhat AS or ES because I don't understand what I'm paying for. *I'm* the Redhat support. I just need something that will receive patches and support for more than one year. The 5 year lifespan of the ES versions is nice, but I've NEVER called Redhat for support. I don't plan to.

I build the kernels for each of the servers. I use vanilla kernel.org source with XFS. We sell 2, 4 and 8-way servers. Am I missing out on anything from the "optimized" Redhat Advanced Server kernels? What are other people in this situation doing?

I think it's confusing because we initially chose Redhat for the accountability aspect of having a corporation behind the distro. Now, I'm not sure who they're targeting. I would imagine that most firms that select Redhat Advanced server and are willing to pay the price (>$1000/license) would have a staff talented enough to support it. So why the mandatory support costs from Redhat?

I don't understand the problem.

I'm estimating that you have 150-200 servers (depending on what RH package you get). If those servers aren't generating revenue, or supporting a business unit that generates revenue, it's time to downsize your datacenter. $350K sounds like a lot of money, but it's all relative to revenue. If it's only 2% of last quarters revenue, then why would you consider making a huge IT change just to save a few bucks. Again, if $350K is really a lot of money for your business to be spending on OS upgrades, than maybe it's time to downsize that datacenter because it's not generating the revenue to justify it's existance.

Dragged kicking & screaming

There are circumstances in which you can be perfectly happy with the free version of Red Hat, but you you are forced for other reasons to purchase a commercial support contract. It happened to me. Imagine a shop where with more than enough in-house expertise to support the free version of Red Hat and deal with its frequent release cycle, and had the full buy-in of management on this arrangement.

The catch; using a commercial piece of software in the mix. In our case, a certain database. Being closed-source and totally non-self-servicable in case of serious problems or bugs, it is imperative to have a support contract for the commercial software. Almost all the RDBMS vendors have now altered/clarified their support policy: they will *not* honor a paid support agreement if you are running the free version of Red Hat underneath their software.

Why this policy exists is a question I will let somebody else speculate about...

There is exactly one major RDBMS vendor I could find that will officially support its software running on the free version of Red Hat (as of April 2003, at least), and that vendor is IBM with their DB/2 product.

Unfortunately, we were too time-constrained to port our system to DB/2, so in the end we caved and paid for Red Hat Enterprise so we could get RDBMS support on our existing platform. To this day we have not called Red Hat tech support once and don't expect to do so, ever. The thousands of dollars we paid covered the 3 minutes of effort the sales guy put in over the phone. Not a bad deal for Red Hat. If I were starting from scratch, knowing about the new support policies from the RDBMS vendors, I would have done the project using DB/2. PostgreSQL would have been an even better choice, except our project required real-time database replication, and PostgreSQL is just now getting to the point where that works well enough.

Buy support from your hardware vendor

If you buy servers from IBM, HP, or Dell (among others), you can just buy support hours - and for a lot less.

Rather than buy the OS for every server, buy the support, and just copy the OS. It's my understanding that this is permitted with RH AW/ES/AS. If you don't need the enterprise version, then don't use it.

-Mark

Debian is the one I picked...

I run a small computer consulting company, one thing I usually do is replace their linksys/dlink/netgear broadband router with a linux box. Although much smaller in scale that what you are talking about, I moved from redhat to debian for the exact same reasons you are talking about. Once RH moved to its frequent .0 releases with one-year updates, I knew I had to pick a different distro. Debian is fast, stable, and compact compared to red hat. The package system kicks major ass as well. I don't need support, the only problems I have ever run into were hardware related (or my own stoopid errors)... I will, however, say that dselect is one nasty mofo of a whatever it is... If I want a new package, I just search the debian site for it, and use apt-get. Like others have said, if the support is what you need, you will have to evaluate whether its cost-effective vs other major unix-y providers. If security updates is what you want, then there are several cheaper alternatives to red hat. Redhat needs to provide an alternative for those who don't want actual support, but do need long term updates for multiple years - otherwise, they will see their piece of the Linux pie shrink. That may be what they want, it may be a simple business decision on their part to make mo money. However, I know many, many people who are ditching red hat for the exact reason I did.

Maybe this is the place

To demonstrate that if a company starts acting like MS it gets treated like MS. I'd take a hard look at the transition costs moving to SuSe. It'll take some testing, no transition is painless. I like Suse Enterprise servers, so I'll admit to some bias. Hey, you have to have standards. When a Linux provider starts acting like MS (forced upgrades, ever escalating prices) maybe it's time to bitch-slap them back into line.

Why do you need to upgade?

This may be a troll, but most enterprises stay with whats working. Why do you feel the need to upgrade all your linux servers? What will a 2.4.22 kernel get you that your not getting with a 2.4.10? Oracle (and DB2) are not kernel dependent.

A new enterprise database? Fine start a new pilot project. Document honestly what you find and make a migration path. Decide on a Win32 migration path to MS SqlServer if you have too. Whatever is best for your company.

Your question demands more information.
Enjoy,

another voice in the crowd

I say go with SuSE Enterprise Server.

It has some nice features like remote auto install and YaST for a very nice system installer and maintainer.

SuSE Enterprise also supports x86, IBM mainframe, I/P series IBM servers, Itanium2 and AMD Opteron processors. This gives you a lot of flexability to add new hardware to the network to improve performance and the knowledge that the new machines will run perfectly with exsisting servers.

SuSE also has great tech support services at a much lower rate than redhat. You can feel confident that your server software is also run by the German Government and praised! by them.

SuSE's max turnaround time for support is just 2 hours!

SuSE is also United Linux Compatible and LSB compliant.

Suse Prices are not too bad either:
x86 single server 749USD$
Itanium Single server 448USD$
Opteron Single CPU 448USD$
Opteron Dual CPU 767USD$
Opteron Quad CPU 1405USD$
Opteron 8 CPU 2585USD$
These include 1 Year Maintainance and Service.

Linux Plus Oracle Support

Support for the particular Linux distro is one thing, Support for Oracle Running on that distro is another thing all together.

At home I run the developer edition of the Oracle 9ias enterprise database as well as release 2 of the Oracle 9ias Application server. I have successfully installed to Red Hat (version 8.0 not Enterprise), Mandrake and also a Debian Distribution. At work we are running a development environment on Red Hat 8.0 and production using Solaris 8. Since we are using pure java and j2ee code our software runs flawlessly across the systems with no changes whatsoever, even considering the fact that some of the developers on the project run Windows systems on their desktops where they actually write some of the source code modules!

If you expect support from Oracle concerning an Oracle installation of any kind whatsoever on Red Hat Linux you best be using Enterprise and yes the support pricing is quite high compared to what most of us are used to running Linux over the years.

I would suggest running your most critical servers on Red Hat Enterprise and if you have supporting environments, perhaps a development or test environment, use Red Hat 8.0 (or even Debian or Mandrake) and save yourself some cash outlay in that way.

If you have the talent within your staff to self support I can attest to the fact that Oracle products run as advertised on Red Hat 7.3 and 8 (have not yet tried 9) using the install procedures Oracle outlines for use on Red Hat Enterprise, but as has already been pointed out, Oracle will not support installations with problems unless you have the Enterprise edition as your underlying Linux Distro.

The Matrix is real... but I'm only visiting!

Re:You can't make copies if you are a customer

In that case, drop RedHat.

Others offer Linux without this problem.

E.g. SuSE allows you to install their enterprise version on as many machines you like. You buy as many licenses as you like. The only limitation is that you get support only for the number of machines that you licensed.

So when you want to save money you just buy licenses for a couple of the most important servers, or for each server of a certain kind.
(e.g. you run 50 fileservers, you license less than that, and when there is a "problem with fileservers" you let them support one of the licensed servers and you apply the fix they supply to all 50 of them.

Make a deal

With that many machines, you'll get better pricing. The organization that I work for (which is huge, but has about 200 linux boxes) pays approximately $200/box.

For us, it was worth it because we are guaranteed a supply of patches & support for a minimum of five years. Red Hat public releases churn every 18 months or so, which is too much work to maintain.

Maybe this is simplistic...

...But why doesn't he just upgrade to RH9 for his Oracle app and run another distro for the other stuff? Certainly, many data centers carry a variety of OS on their servers... Even in our (mostly) Windows back room, we have a couple Solaris machines and a couple Linux boxes and everything works out just fine. (Except for the Windows stuff which present constant niggling, unexplained/undocumented problems. Exchange server? What a nightmare. Thank god it isn't the actual external mail server or the real relay.)

Anyway, this is a bummer, because I certainly do appreciate what RedHat has done for the comunity. I know MY first box ran RedHat. It is too bad the prices are so high, but if you run a business critical app, it is almost neccessary to have 24x7 support available. Even though you'll use it two or three times in your whole life, when you need it, your ass will be saved. I've called MS support one time and only did so after a lot of work...But finally, we had no choice. It was get the thing up or we're out of business. OK...We'll call Microsoft. They did suggest a lot of things we had already tried, but ultimately did help us. We had to pay something like $300 for the call, in addition to our enormous annual Microsofft budget, but we got it running.

My urge is to move everything to something else, but our CIO lives in the dark ages and won't listen to any suggestions that aren't Windows...Our two linux boxes function as load balancers for...IIS. They exist because our Win NT 4.0 balancers were choking and we didn't want to spend $25,000 on new Windows hardware to cover the same task. I weep because I know that the two balancers would serve the site better than the six win2k boxes which do the same job now. The Sun boxes were installed with the phone system. Same weekend, we put the Exchange 5.5 server in. You can guess which has gone down more times since.

Right now it is Exchange seven crashes, Sun ZERO. Not even a reboot in the last 12 months! And the last one was because of a wide-scale, longterm power failure that outlasted our UPS and generation capability.

How I see it

For any distribution/os to be acceptable it needs the following:

• Support: This means email, phone and on-site if need-be. You can eliminate the last two if you are already employing very good administrators. However management often wants "piece of mind" or in legalese "due diligence" for their shareholders. This means being able to show that they have ensured that if something goes wrong, they have someone to go to.
  
• Fewer product cycles: You do not want to have to upgrade your operating system and/or hardware more than every 3+ years. Anything below that is too costly if you have a large number of servers.
• Patches - fast and simple to apply: In this arena, Solaris is pretty good. RedHat and Linux distributions in general seem to suffer from dependency lunacy. That could be just me though. FreeBSD and the BSD's unfortunately are atrocious in this area at the moment. To install a bugfix you have to have your own cvs/build server in most instances - there are no official binary patches available. If you have skilled admins it isn't a problem.
• Commercial application support: Out of the free O/S's out there, both RedHat and SuSE seem to be the leaders. Of course, you may not need to run for example - 10 different Oracle servers. So you could have 1-3 (production/development) Oracle servers, and run alternative platforms on the rest of your boxes. This "hybrid" architecture would bring your costs down.
• Education/training: Out of the free Unixes, RedHat definitely appears to be the king. Then again, if you have skilled Unix admins who know how to view processes across both SysV and BSD systems, or remove files with a - in front of them, they probably don't need it. Management may however want it for that annoying "due diligence" thing.

So I haven't really given you any answers, but I hope I've offered some good points to consider.

As a Univ sysadmin...

...responsible for several Redhat servers and workstations, I can tell you that we're beginning the process of switching to Debian.

Stable (if some of the software is a little outdated), easy to maintain and upgrade, no registration required to use apt (see how far you get on RHN without that), and far cheaper.

Look at Sun

Currently I am working with a lot of Sun kit - and their sales guys. They are absolutely thrilled with Red Hat's new pricing, becuase suddenly they are competitive again.

Consider - a small Oracle DB on a 2 processor machine. The cost of a decent 2 processor server is about $2000, and then the cost of RHAS is about $2700 as I recall. Suddenly the cost of a V240r doesn't seem that bad. We pick them up for a lot less than $4700. Of course we have a pretty good deal with Sun, and the poster may get a good deal with Redhat, but we've done the analysis, and RH does not stack up for us in this example. For me, in is interesting that we have said "no Linux", not because it is a "hacker OS" or it can't do the job - but because it is too expensive to deploy. And before anyone asks, we didn't do any TCO voodoo to prove the point!

Other things Sun have on their side:

- Scalability on the same architecture. Yep, I know 2.6 will scale, but it isn't even properly released yet. We develop on small machines (240s, 480s) and deploy on 15Ks without even thinking about it - apart from making sure that the app can use the CPUs
- Solaris - damn good OS, excellent support and an understanding of what enterprise computing is about
- Support. Judging by some of the comments here Redhat support is somewhat lacking. Having called regularly on Sun support, I can say it is quite exceptional - even when problems are not their fault, they will engage with other vendors to get a fix

Red Hat in the Enterprise

At my last gig I was responsible for ~30 Linux servers, all running Red Hat. There were about 5 of them running RHAS 2.1 and the rest were running 7.x

I spent a couple of days with the Oracle DBA benchmarking our applications and found it interesting that 7.3 was a tad faster than RHAS 2.1. Hardware was IBM x345, dual 2.4GHz Xeons, 2.5GB RAM, ~200GB RAID 0+1. Yes, hyperthreading was disabled.

I find it odd that Red Hat's "Enterprise Linux" is missing some key enterprise features that can be found in its consumer distribution (such as Logical Volume Management). BTW, LVM is broken in Red Hat unless you compile your own kernel, otherwise you can't mount snapshots.

In any case, Red Hat's new pricing scheme is flat out extortion. I had enterprise support on my servers and ever single time I reported a problem, I was either delayed until I found the solution for myself online or flat out told "That's not supported." You might wonder what's not supported. How about LDAP authentication? The automounter?

There are some things about Red Hat that are wonderful. And some that are pretty good. In the wonderful category, their installer is just the bees knees. Especially if you're kickstarting your servers. RHN is a nice tool but fundamentally flawed in that you must use Red Hat's repository; imagine 30 servers downloading the same 45MB of RPM's over a T1 at the same time over https (which can't be cached). yum goes a long way towards fixing this.

Debian has some nice points but IMHO has a lousy installer and zero enterprise management tools (such as RHN for Red Hat). People have been bitching about the installer forever and nothing seems to be getting overhauled there.

If I were doing this from the ground up right now, I'd go with RH 9. Keep your eyes open and keep track of major releases by RH and evaluate for yourself when an upgrade is necessary. RHEL is made up of many components that have been deprecated in the mainstream release (such as CUPS, I think Sendmail may also be deprecated). For LVM features you need >8.0 anyway. Use yum for your package management, build your own local package repository, and spend a little time learning about the guts of RPM.