Oracle's Take On Red Hat Linux

darthcamaro writes "For nearly three years, Oracle has had its own version of Red Hat Enterprise Linux, claiming the two versions are essentially the same thing. But are they really? As it turns out, there are a few things on which Oracle and Red Hat do not see eye-to-eye, including file systems and virtualization. The article quotes Wim Coekaerts, Oracle's director of Linux engineering, saying, 'A lot of people think Oracle is doing Enterprise Linux as just basically a rip off of Red Hat but that's not what this is about. ... This is about a support program, and wanting to offer quality Linux OS support to customers that need it. The Linux distribution part is there just to make sure people can get a freely available Linux operating system that is fully supported.'"

Comments

Total Flamebait

[phantomfive]

This is great. When the only thing differentiating Redhat from Oracle is service, Redhat will win because they are the ones actually creating the product. If there is any single company that I'd like to see pushed into the ground by open source, it is Oracle. Whereas Microsoft is kind of a bumbling giant that can't quite get things right but gets by on chair throwing, Oracle is downright evil. They will actively destroy another company if it makes them a cent.

On the other hand, Oracle is much less likely to go under because they produce other things of value that the open source community will have difficulty replacing (because we don't do much business software).

Re:Total Flamebait

[aralin]

This has nothing to do with who makes the product. It is entirely a support issue. Lets say you have a problem with your Oracle solution on Solaris or RedHat Linux. Oracle will look at the problem and determine it is an OS problem and so you go to Sun or RedHat and they say this is after all not an OS problem, but a virtualization problem so you go to third provider, who will find out that finally fixes the problem two day and several millions in lost profit later.

If you can have one provider who will offer support for the entire stack, OS, virtualization, database or middleware engine, you have a huge win on your hands. Premium contracts can have time limits which now don't cover just one layer, but the entire stack. The same company will resolve the problem no matter where it lies and they are responsible by the service contract to resolve the problem. Where the problem actually lies is an internal issue you don't need to care about.

When you add to it that business talks are done with single company, which results in time savings and you usually save by bundling the service contracts into one package as well. this is almost a no brainer that customer actually demand this.

Add to it that RedHat is not binary free product, that you actually have to pay for the binary distribution of enterprise version, and that Oracle will basically save you additional money by compiling RedHat linux from sources for you.

Redhat has a huge edge for servers not using Oracle database or middleware, but for servers actually running Oracle products, it is no brainer to go with a full stack support contract.

Re:Total Flamebait

[Anonymous Coward]

It's also a very smart business move by Oracle. Pushing free operating systems running on commodity hardware allows Oracle to reduce the price of an Oracle based solution without reducing Oracle's revenue. That's business savvy.

Re:Total Flamebait

[Anonymous Coward]

Have you ever actually used RedHat support? Judging from my experience with their support quality, it seems they are only interested in supporting new installation problems for small businesses and thats it.

Traditional high-end enterprise support from Sun and Oracle makes Red Hat look like the "small time" shop they are.

Re:Total Flamebait

[wireloose]

In my job, I am supporting both versions, RedHat and "Oracle." I get better support from RedHat. It's enough better that I have gotten away with converting several of the "Oracle" servers over to RedHat. There are still a few, but a few less every year.

Re:Total Flamebait

[ebuck]

Misinformation extradorinaire!

RedHat is binary free. You can download any current distribution of RedHat compiled. With some hunting around, you might even find the old ones. What you can't expect is for them to make available the binaries for the patches and bugfixes for that distribution.

Without the binaries for the patches and bugfixes, it is still binary free. You just have to use the rpm build system and do a lot of rpmbuild --rebuild to get the binaries. Failure to do so on your part is not lack of freedom, it's laziness.

RedHat doesn't point the finger at someone else and say "We can't fix it due to him!" There is a reason RedHat supplies so much back to the open source community and a lot of it has to do with supporting their customers.

From both of your statements, I'd wager that you've never paid for a RedHat product; or if you did, you never bothered to register it.

Oracle wants an operating system that's specifically tuned for it's product. That's great if all you intend to do is run Oracle on it, but most people eventually use their system for something else too. Then the tuning doesn't pay off, it works against you.

I remember the earliest Oracle installations, there were so many tweaks and configuration changes to the kernel, shared memory, enabled extensions, ... it seemed to go on forever. If Oracle had it's way, I'd imagine that they would eventually turn a PC into an embedded Oracle server. That's not necessarily a bad thing, unless you are in the business of selling full operating systems, like RedHat is.

Perhaps Oracle's treatment of Linux has changed a bit, but I'm not holding my breath. Old dogs use tried-but-true old tricks. Since Oracle still hasn't seen fit to rewrite their init.d stuff to be half sane, I doubt they are concerned about being friendly to others on a Linux server.

Re:Total Flamebait

[reiisi]

Well, you know, if I start wanting to do something else with the hardware, I can always install RedHat. (And probably migrate to postgresql.)

The one thing that bugs me about Oracle's customization of RedHat is the question of whether they are giving back, both to RedHat and to the community. Maybe I don't look in the right places to know, but it sure isn't obvious that they do.

Actually, I'll go a little further and put it this way: From a potential customer's point of view, if I'm going to dedicate a lot of my infrastructure budget to Oracle's products, I would be a lot more comfortable if I knew that Oracle and RedHat had a good working relationship, both technically and economically.

(Cannibalizing the market is really not in a company's own best interests, any more than relying on the schoolyard bully to protect you is.)

Re:Total Flamebait

[reiisi]

No it isn't smart. It's just very short-sighted, the same way that lots of "smart" things being done in financials the last ten years were just very short-sighted.

You get what you give.

Re:Total Flamebait

[MoreDruid]

Yes you get support from 1 vendor, however I recently had to deal with that vendor. There was already a support case, all kinds of log files uploaded for them to analyse and after 1,5 week they hadn't found the issue yet. What was the problem? at 4:20 every night 1 of the servers in a cluster of 4 went down. The issue was that updatedb was configured to run on OCFS filesystems, and updatedb is triggered by cron.daily. They had about 5 different engineers looking at our case. No solution, until someone from my company decided to dig a little further into the updatedb config. It seems that you shouldn't run updatedb on OCFS filesystems (we have another customer who has been doing that for more than half a year with way more nodes concurrently connected, but hey). Note: this happened after issuing a Oracle CRS (cluster software) update, the config had been running fine for more than a year. And Oracle support just kept on looking to the Oracle part, ignoring the OS stack. From Oracle Applications support I was told to "just update glibc from 3.2 to 4.x because there's a bug that's fixed in 3.6". Right. Break compatibility with all your major tooling and applications so you can run an Oracle App because they've been too lazy to test in an "old" environment (RHEL 4 U4).

In short: I'd rather deal with 2 or 3 independent vendors who know their shit (and know it well), than with 1 vendor who would - even when told differently - kept looking from the wrong POV.

Re:Total Flamebait

[BestNicksRTaken]

I gotta say thats what I've always found with paid-for support, if you want the job done right, do it yourself.

Just the other week I had a problem with a certain piece of software we had bought to pentest a pretty dated operating system, after about 3 weeks of uploading logs and changing configs and turning on debugging for the "tech support" at both the software vendor and the OS vendor, I decided to have an in-depth look at the problem.

Guess what? Identified and solved the problem in about an hour.

And as far as Oracle goes, they should just make all their bloody shared memory and kernel tuning "enhancements" (that tend to break everything but Oracle) into a shell script or RPM and leave RHEL alone, its not as if the changes they make are enough to warrant a complete OS rebuild; they're doing it just to stick it to RedHat as there are some PHB's and DBA's who actually will use Oracle Unbreakable.

Re:Total Flamebait

[jabithew]

Explain? I can't decide if your post is cryptically insightful or just content-free aside from a vague, zeitgeist-ish, anti-corporate rant.

Re:Total Flamebait

[msormune]

You're right about the Oracle proving really something open source community can't do much. They're called "jobs".

Re:Total Flamebait

[DrgnDancer]

I'm with sibling on this one. I can't figure out if you've seen some clever angle I missed or just spouting baseless drivel. As far as I can see this is a pretty smart move by Oracle. They are providing a standard OS that they know their product will run well on, using something that a lot of sys admins are familiar with to do it, and adding only minimally to the cost of their product to do it. Seems like a win all around to me.

Re:Total Flamebait

[sulfide]

HAHAHA I work for one large enterprise that was recently swallowed by another.. Oracles support is a joke, 95% of the time they provide nothing useful other than drones that sit on the conference call hoping that someone else figures out the problem first. Support is just over paid warm fuzzies, and I'd rather see that fuzzy money go to redhat who does something useful with it, than larry's billion dollar estates.

Re:Total Flamebait

[Anonymous Coward]

If your database is small enough that you'd consider running something else on the same server then you don't need Oracle.

Frankly, Oracle is fucking horrible. The only time you should ever use it is when you need to do something only Oracle can do. Which isn't running small databases on shared servers.

Re:Total Flamebait

[lawpoop]

at 4:20 every night 1 of the servers in a cluster of 4 went down.

I think I found your problem. You might do a 'random' drug test of the server room.

Re:Total Flamebait

[lawpoop]

In short: I'd rather deal with 2 or 3 independent vendors who know their shit (and know it well), than with 1 vendor who would - even when told differently - kept looking from the wrong POV.

You really think you're gonna get hooked up with three 'knowledgeable' vendors who know their shit and are willing to go to bat for each other, or are you more likely to get the same level of crappy support from 3 or 4, with hot potato tossing added into the mix?

Re:Total Flamebait

[reiisi]

I ranted a bit more completely elsewhere, but I'm basically saying that I don't think Oracle is doing this the right way.

I am reminded of this.

Re:Total Flamebait

[TheEnlightenedOne]

On the other hand, Oracle is much less likely to go under because they produce other things of value that the open source community will have difficulty replacing (because we don't do much business software).

What about postgreSQL?

Re:Total Flamebait

[phantomfive]

PostgreSQL is an awesome piece of software, but database is only on part of what Oracle does. If you go to their website, you will see a ton of software, a lot of which I only have a vague idea of what it does, but it's for various business purposes.

I had a friend who worked with Oracle's peoplesoft system, and he said it would take several years to reproduce the whole system they have, templates and everything. I am not sure what it is really either, but apparently it saved his company a lot of time and money. He hated it, though.

Re:Total Flamebait

[alexborges]

Postgres+Jboss

Re:Total Flamebait

[Octopuz]

The problem with Oracle Support is that an issue is handled by a single product support team - and I have been part of such a team. The team is specialized in its own stack and can hardly look at the other side of a wall. There are two walls to throw the service request over: the first wall is towards a different team (which will defend the wall against intruding service requests), the other wall is a different timezone. Urgent SRs may need to travel across the globe. That means another engineer will start from scratch investigating the problem. This all does not contribute to a quick resolution process.

Re:Total Flamebait

[BitZtream]

What a gods damn minute! You mean Oracle told you to upgrade to a version of the software without the bug in it?! Those fucking bastards. The should have fixed it completely without changing anything, even a configuration file!

Seriously if this bothers you, you need to not deal with F/OSS software. Backwards compatibility is no where on the feature list, its almost 'broken by design' as those developing the code have no need or reason personally to maintain backwards compatibility. You have the source, you can fix the bug and recompile!

Sounds like they should have found it quicker, but acting all surprised because they wanted you to perform a system wide change that would effect many if not all applications. Why are you running Oracle on a system with a bunch of other stuff on it? If you can afford to run Oracle and a cluster at that, it would seem to me that you can probably afford to put the DB servers on their own systems. Thats what pretty much everyone else does who has been admiring for more than a month. At least it should be dedicated to its own jail host (or whatever your OS wants to call them). Application compatibility issues like this are something every major unix is designed to deal with.

Perhaps the Oracle engineers just gave you too much credit as far as your abilities? It really shouldn't have been that big of a deal compared to your crashing.

Re:Total Flamebait

[afidel]

For what you pay per core for Oracle licensing you would have to be an absolute idiot to run anything other than Oracle and possibly a backup application (though we don't) on your Oracle server. Having a tuned version of Linux specific to Oracle is a great idea, the problem comes when you get into the reality of support. Perhaps we aren't a big enough fish to matter to Oracle but our experience has been some of the worst support I have seen from ANY organization. That is a large reason why I'll be strongly suggesting Redhat for our Linux support later this year despite the fact that all we will be running on it is Oracle.

Re:Total Flamebait

[afidel]

Urgent SRs may need to travel across the globe. That means another engineer will start from scratch investigating the problem.

Then their follow the sun support system is broken, pure and simple. When I work with Microsoft or Cisco I don't have to retread the entire process every time a new engineer joins the team, the vendors have internal mechanisms to bring new engineers up to speed on the case. Only when an engineer wants to provide a new perspective on something we have already covered do I get asked to rehash something. Oracle on the other hand seems to always require going back over things from square one, whether it be follow the sun support or re-opening or updating an existing ticket.

Re:Total Flamebait

[partenon]

Seriously if this bothers you, you need to not deal with F/OSS software. Backwards compatibility is no where on the feature list, its almost 'broken by design' as those developing the code have no need or reason personally to maintain backwards compatibility. You have the source, you can fix the bug and recompile!

Backwards compatibility is one thing. Broken stuff is another. If something is broken, the Linux Vendors will fix it (depending on their knowledge). But if you updated software X to version 2.0, you should be aware that it may require a newer version of lib Y.

cant we already get free and support with centos?

[doktorjayd]

i've never bothered to look at oracle linux, because i can get 'free' redhat through centos, and when i want paid support, i can get it directly through redhat.

without some other differentiation, what is oracle providing that isnt there from the others?

so yes, it is just a rip off of red hat.

ORACLE

[fuzzyfuzzyfungus]

One Rich Asshole Called Larry Ellison...

Re:ORACLE

[robmv]

Or reading it backwards "EL CARO", translated from spanish "The Expensive"

Re:ORACLE

[leamanc]

I always thought it was "Operation Release All Cash to Larry Ellison."

Re:ORACLE

[AJWM]

Not so much any more, but in the early seasons of Smallville, Lionel Luthor kept reminding me of Larry Ellison, only with longer hair.

Re:ORACLE

[init100]

You can also substitute Rich with Raging.

Um,

[Darkness404]

Um, so basically it is a rip off of Red Hat just with Red Hat stripped out and Oracle's own filesystem added to the kernel, with a different VM. Thats it. Still maintains binary compatibility, etc. This is basically like someone trying to justify that Linux Mint is some grand new distribution when it is nothing more then Ubuntu with a few extra tweaks and drivers added.

Re:Um,

[Anonymous Coward]

Um, so basically it is a rip off of Red Hat just with Red Hat stripped out and Oracle's own filesystem added to the kernel

No no no no no.

The default filesystem shipped with RHEL and OUL is ext3. The clustered file-system shipped with RHEL is GFS, and with OUL it is OCFS2. OCFS2 is not compiled in-kernel within Red Hat Enterprise Linux, and must be mod-probed in as a separate - unsupported by Red Hat - module.

> with a different VM.

Again, wrong. RHEL 5 ships with Xen, and will support Xen until at least 2014. OUL also ships with Xen. Please remember, KVM has not shipped in *any* RHEL release (major or minor) yet. Only Red Hat internally knows the release agenda.

> Thats it. Still maintains binary compatibility, > etc.

Oracle's binary compatibility claims are a myth.

Re:Um,

[chelsel]

Hey, this is open source... if Oracle is doing something permitted by the license agreement then what is the problem... if it's not "in the spirit" of open source then maybe it makes sense to update the license agreement.

Re:Um,

[1s44c]

Please remember, KVM has not shipped in *any* RHEL release (major or minor) yet. Only Red Hat internally knows the release agenda.

Apparently they will release their KVM based vitalization before the management tools run on anything other than Windows 2003. That's what they got from Qumranet and that's why I won't be using it.

KVM is going to be slower than XEN unless you have a super-duper-mega-new CPU with Intel EPT or AMD RVI support.

Re:Um,

[Darkness404]

The article though tried to make it sound like Oracle's OS was so much more improved then what Red Hat had because they changed the kernel and added a VM but nothing more.

Re:Um,

[Ant P.]

This is basically like someone trying to justify that Linux Mint is some grand new distribution when it is nothing more then Ubuntu with a few extra tweaks and drivers added.

Which, in itself, is a lot like someone trying to imply that Ubuntu is a distribution when it's nothing more than a snapshot of Debian sid with a few extra tweaks and drivers added.

Re:Um,

[Mad Merlin]

Apparently they will release their KVM based vitalization before the management tools run on anything other than Windows 2003. That's what they got from Qumranet and that's why I won't be using it.

[ citation needed ]

Seriously, I don't even think KVM runs on Windows at all.

KVM is going to be slower than XEN unless you have a super-duper-mega-new CPU with Intel EPT or AMD RVI support.

You mean like any CPU either have released in oh... about the past 3 years?

Re:Um,

[Anonymous Coward]

> Seriously, I don't even think KVM runs on Windows
> at all.

The OP is correct. They are referring to RHEV for Servers: http://www.redhat.com/about/news/prarchive/2009/agenda.html

Re:Um,

[Anonymous Coward]

Only Red Hat internally knows the release agenda.

In the upcoming RHEL5.4 KVM will already available next to Xen.

Re:Um,

[Spit]

Oracle is following the spirit of open-source: taking the source and doing something else with it. Someone could totally take the Oracle distro, patch in whatever mods they like and call it "FUCK ORACLE LINUX".

Re:Um,

[BitZtream]

If you mean like every other Linux distro in existence, then yes.

Redhat is a rip off of those who came before. The nature of GPL and is why there are so many Linux distros.

How this got modded insightful is beyond me. They are doing the exact same thing every other Linux distro does. Take someone elses as reference, repackage/modify it to suit them, release.

They are ripping off Slackware, which ripped off SLS, which probably ripped off some other distro before it. Thats the way GPL'd software is SUPPOSED to work.

Bunch of retarded hippies, 'you must share our software but you can do anything you want to it otherwise cause we're for FREEDOM!!!' and then when someone uses it 'oh my god, you're stealing our freedom and software because we didn't think of restricting that freedom in the current revision of the license!!!! We'll have to remove that restrict that freedom so that it can be more free!!!@$!@$twelve'

Wake up and smell the coffee, all modern distros are rip offs of someone else. Whats good for the goose is good for the gander, stop whining and put your thumb back in your mouth.

Re:Um,

[1s44c]

[ citation needed ]

Seriously, I don't even think KVM runs on Windows at all.

I never said it did. The RHEV management console runs as a .net application on windows 2003. VMware ESXi and the commercial xen offering both also require windows consoles. It's like some sick joke the virtulization work is playing on us.

KVM is going to be slower than XEN unless you have a super-duper-mega-new CPU with Intel EPT or AMD RVI support.

You mean like any CPU either have released in oh... about the past 3 years?

Ok, say EPT/RVI has been in every chip produced for the last 3 years. Do you seriously expect industry to change all its servers within 3 years? Or throw out servers that work perfectly well with Xen only to spend money buying something new that runs at the same speed. Responsible IT is about staying off the upgrade treadmill.

Re:cant we already get free and support with cento

[NaCh0]

Oracle Cluster File System. Whether you need it or not is up to you. Oracle also provides OCFS modules for Red Hat to make it easy on people.

Re:cant we already get free and support with cento

[Anonymous Coward]

i've never bothered to look at oracle linux, because i can get 'free' redhat through centos, and when i want paid support, i can get it directly through redhat.

without some other differentiation, what is oracle providing that isnt there from the others?

so yes, it is just a rip off of red hat.

Um, no you can't. Red Hat does not support CentOS.

PS - it's 'Red Hat'.

Re:cant we already get free and support with cento

[Wakk013]

There is a free version of Oracle available. Its not licensed in any way. However if you go over their max data limit, then you have to purchase a version of Oracle.
Also, RedHat and CentOS are the same product. They are the same source code: RedHat compiled by RedHat, and CentOS compiled by open source community. This allows RedHat to get more exposure and most of the bugs found in CentOS can be patched back into RedHat.

Is anyone actually using Oracle Linux?

[lacourem]

Serious question. My employer has recently stated that they would prefer us to use Oracle Linux for future installations instead of Red Hat. Just looking for some insight from someone else who has taken the plunge.

Re:Is anyone actually using Oracle Linux?

[EarlW]

I've installed a few UL Oracle systems. I wanted to be 110% sure that they would run the Oracle database. They just changed 'Red Hat' to 'Oracle'. They run fine...

Re:Is anyone actually using Oracle Linux?

[mrphoton]

I would not take the plunge. Just grep the kernel source, how many times does the word oracle come up compared to redhat? Redhat and the people they employ have been responsible for a tone of linux development - it is their core business. Oracle have comparatively done bugger all. Redhat employ _real_ kernel developers. Do oracle? So when your server crashes in the middle of the night, who do you think will be able to produce a kernel patch to fix the problem, the people who employ kernel developers and have done so for years. Or.... a company that three years ago decided to rip off somebody elses distro... I would not touch oracle with a barge pole. (disclaimer: but all that is only opinion)

Re:Is anyone actually using Oracle Linux?

[Anonymous Coward]

Serious question. My employer has recently stated that they would prefer us to use Oracle Linux for future installations instead of Red Hat. Just looking for some insight from someone else who has taken the plunge.

What about any other third-party applications?

If you're using SAP, JBoss, etc., and they're only certified to RHEL (and not Oracle Linux), then there's no sense having two different distributions. Try to keep things as cookie-cutter as possible to minimize variables you have to worry about.

Re:Is anyone actually using Oracle Linux?

[eln]

Those guys can be really picky too. You can tell them your Oracle Linux or CentOS stuff is identical to RHEL, but if it doesn't actually say RHEL on it, they won't support their stuff at all.

My advice is to use Oracle Linux and take advantage of their end to end support contracts for the servers that are running Oracle products (and keep in mind Oracle owns a lot of products businesses use now like Peoplesoft, JD Edwards, Siebel, etc), and run RHEL on servers that are running non-Oracle stuff.

Re:Is anyone actually using Oracle Linux?

[carlzum]

My company has been moving from Solaris to Oracle Linux recently. It's been largely unnoticed by the users and DBAs. The decision to go with Oracle Linux was a cost savings measure, but we also like the idea of Oracle supporting OS-related problems. Red Hat may provide better support than Oracle, but we've spent a lot of time eliminating potential OS issues working with Oracle support.

So far, we haven't run into anything that required Oracle's help with the OS. IMO, Oracle Linux is a good option if it supports the features you need, offers a lower TOC than the alternatives, and your admins are already familiar with RHEL. Running Oracle on RHEL would have cost an extra $300-400 per server under my company's licensing agreements. Depending on the number of servers you have, the cost difference between the two may be negligible.

Re:Is anyone actually using Oracle Linux?

[j-cloth]

I tried to do the same thing but found, at least when OEL was first released, that the Oracle software stack (particularly Oracle Applications) was better supported on Red Hat than on Oracle's distro.

Now, they've caught up and I have some of both types lying around. To be honest, it doesn't really matter to me which is which and I don't think I could actually tell you which distro a particular machine is running without checking /etc/redhat-release.

Re:Is anyone actually using Oracle Linux?

[Anonymous Coward]

My company has been moving from Solaris to Oracle Linux recently.

Sad :(

grep wtfismystorage /proc/scsi/scsi; nohup stabselfinface &

Re:Is anyone actually using Oracle Linux?

[reiisi]

This is one of the reasons I'd prefer Oracle to have a good, cooperative relationship with Red Hat, rather than a parasitic relationship.

They could be working together to keep the APIs the same.

I will admit, however, that where I might use an Oracle server, I'm most likely to be accessing that server through Oracle. On the other hand, there is peace of mind in the sense of being able to fix it yourself if you know that SAP, JBoss, etc. can get around Oracle if necessary.
 

Re:Is anyone actually using Oracle Linux?

[Anonymous Coward]

Oracle have comparatively done bugger all. Redhat employ _real_ kernel developers. Do oracle?

Yes. http://kerneltrap.org/node/7637 as one example.

Re:Is anyone actually using Oracle Linux?

[Macka]

"lsb_release -a" is a better command, because it works on all major distributions, regardless of the type.

Some reasons for the Oracle case

[DiegoBravo]

I support a software product in a telco, and had talks with its IT managers about the Oracle Linux issue. They have lots of Red Hats but see the Oracle offering interesting (and are implementing it) because:

1) Linux (RedHat or others) are really stable systems (compared with other Unixes they had or have), so the support provider switch is not seen as a dramatic issue
2) They can save some cents without (apparently) giving anything. The RedHat support is little money for that kind of company, but a saving always looks good for the directors
3) They avoid one provider's negotiation as a whole (which is a big win: less paper, less meetings, less vendor talk, less decision process, etc.)
4) They mostly ignore the distributed filesystem issues, and for virtualization just apply the leader (VmWare), so the Xen/KVM/Xen-Oracle discussion is not too relevant
5) BTW, for some diverse reasons, their software providers seem to dislike CentOS (maybe the RedHat's negative marketing made its effect, who knows)

Re:Some reasons for the Oracle case

[Anonymous Coward]

1) Linux (RedHat or others) are really stable systems (compared with other Unixes they had or have)

Really? What might these other "Unixes" be?

Linux has gotten fairly stable in the last couple years, but I don't really think all that many people would hold it up as the gold standard.

I have Solaris boxes that have been up since before the 2.6 Linux kernel was released.

Re:Some reasons for the Oracle case

[McPierce]

Not sure what you mean with #5, but there's no negative marketing WRT CentOS by Red Hat. CentOS is friends with Red Hat and are symbiotic with them: they provide to the market segment that can't or won't pay for support, and when people ask CentOS for support they're directed to Red Hat.

NOTE: I work for Red Hat and we discuss CentOS from time to time internally. It's a generally well accepted relationship we share with the folks at CentOS.

Re:Some reasons for the Oracle case

[seifried]

CentOS is really, really slow on security updates and updates in general.

Re:Some reasons for the Oracle case

[Anonymous Coward]

CentOS is friends with Red Hat and are symbiotic with them
NOTE: I work for Red Hat and we discuss CentOS from time to time internally. It's a generally well accepted relationship we share with the folks at CentOS.

Perhaps you can answer something I've wondered about: Why does Red Hat no longer allow Centos to use the words "Red Hat" on their home page? They've been replaced with awkward phrases like "a prominent North American Enterprise Linux vendor" and "upstream vendor".

Re:Some reasons for the Oracle case

[Anonymous Coward]

Because the words you quote are trademarks of Red Hat, Inc. You would have to ask Red Hat's legal counsel why they are unwilling to allow them to be used. Personally, i find it quite obvious.

Re:Some reasons for the Oracle case

[carnalforge]

3) They avoid one provider's negotiation as a whole (which is a big win: less paper, less meetings, less vendor talk, less decision process, etc.)

Then sure as hell they're not the big companies where i've been working as external. Hell, seems that's what seems they enjoy most, 4 meetings just for 1 shell script's line modification with people from 3 other companies, talk, talk, talk.

Re:Some reasons for the Oracle case

[Anonymous Coward]

Since when can't trademarked words be used by the public or competitors? Some competing products even have "compare to YYY" on their labels or advertisements! I was wondering why Centos feels they can't even say "Red Hat" on their front page.

ooh the controversy

[mutantSushi]

I'm not THE most knowledgeable on the minutae of these, but all the bad blood about Red Hat/ Oracle seems silly: The whole point of "Free Software"/Linux is that any company does not "own" code or software (well, they still do, but give up any claim to interfere with others' use of it). Commercial Linux companies obviously need to make their money thru support services. So Oracle thinks they can compete against Red Hat in this area. Obviously, Red Hat as the signifigant maintainer/updater MAY have an advantage. All the end-users get to decide it themselves, and since the code-base is so close, it's relatively easy to switch back and forth. What is the problem when "Free Software" is working exactly how it's supposed to? So what if Oracle eats Red Hat's business for lunch without contributing back? Linux will still be improved by those who want to improve it. All that such a scenario would mean is that (if it occurs) the model of maintenance/support service subsidizing development may not work for all cases. If that's true, then so what?

Re:ooh the controversy

[aralin]

Oracle is essentially a parazite on the Redhat Linux. It is not in Oracle self-interest to eat RedHat's lunch because if RedHat goes bust, Oracle won't be able to maintain it by itself. Oracle only wants to eat the minimum amount of RedHat's lunch needed for its own profits, which lie in selling their own products with full support of the entire vertical stack. So clearly Oracle will go for support contracts on servers that run Oracle Database or Middleware solutions, or maybe as part of an existing contract few other servers that are essentially part of the entire solution, but which might run other products. Oracle would be against itself to try to compete with RedHat in other areas and make RedHat's business model financially unviable.

Re:ooh the controversy

[pembo13]

The controversy is that Oracle is quite literally a leech. And is trying to take money away from a company that is heavily invested in the Linux community.

Of course RedHat's failure wouldn't kill the community, but it sure wouldn't help it. And every penny RedHat looses is one less penny to hire a FOSS developer.

Quality Support?

[StormReaver]

Oracle can't even give quality support for its own software. Why on Earth would it think it can give quality support for someone else's software?

Re:Quality Support?

[El_Oscuro]

Tried Oracle Linux a few years ago. Now, I am a customer of Oracle, what might I perhaps want to use it for? Perhaps to run an Oracle database?

Apparently not to Oracle. Getting the installer to work involved wading through various technotes, installing obscure RPMs, dependency hell, etc. Installing it on Ubuntu was actually a lot easier, and it isn't even supported.

And don't get me started about support. Oracle Metalink now requires flash to work at all. I mean, it is sort of like using GIMP to edit text files.

Re:Quality Support?

[eln]

Oracle Metalink [oracle.com] now requires flash to work at all. I mean, it is sort of like using GIMP to edit text files.

Not sure about your other claims, but this one is just a flat out lie. Right on the front page of Metalink is a selector where you can choose to login to the "classic" HTML-only site instead of using the whizbang Flash version.

Re:Quality Support?

[PowerKe]

Then again, check the news section after logging on to Metalink:

Later this year, Oracle will retire the Classic MetaLink support portal and provide a single support interface through My Oracle Support.

So it does appear that it won't take too long before you'll be unable to use Metalink without flash.

Red Hat Linux has been discontinued since 2003

[Drinking Bleach]

Do they really need to attempt to compete with it?

Oracle understands business

[girlintraining]

Before everyone goes all stupid crazy about Oracle versus Red Hat Steel Cage Match, I'd just like to point out that Oracle has been around since 1977. Redhat: 1995. Redhat brought in $400 million in revenue in 2007. Oracle? $22.43 billion. I could go on, but I think you get the point. Oracle is a freight train, and Redhat is a skinny guy who jogs a couple times a week in the business world.

If I go to senior management and say I'd like to use Redhat Linux, they'll go "What's that?" If I say I want to run Oracle Linux, they'll ask "How much will that save us?" There is no question of Oracle's reliability, or market performance. None. Oracle doesn't need to prove itself. So if you're a fan of getting Linux into the business, you should be saying "hip-hip hoooray" to this; You've got a free pass now at the executive board meetings to install Linux now somewhere. Or... or you can bitch about how it's the wrong flavor of linux and tear into Oracle for ruining the good name of Linux, how Orthodox Linux users are into shaming other users, and Oracle is more like New Evangelical Linux -- half the guilt, twice the usability, etc., etc.

Your call.

Re:Oracle understands business

[pembo13]

Yes, obviously because your senior management has heard of it, it must be what's good for the community, no actual evaluation and long term planning is necessary.

Re:Oracle understands business

[See Attached]

Oracle does need to prove itself. Its got its own tech stacks that seem to be completely independant of each other. How many copies of the JVM are in an app server? Why are versions 8.0.6 still embedded inside the higher versions? Redhat provides coherency, and transparent success. How often does one read "Redhat 5.0 or newer"? Oracle provides the specific checklist of which versions are supposed to work with other versions. Redhat does a much better job at a stable and well-thought out product. Should we expect Oracle to fork Redhat or just pimp it as their own each time a new version comes out? I can't honestly pay Oracle for support when Redhat provides the coherency that leads to success. We have enough would-be visionaries that merely resell products that are "hammered to shape, filed to blend, painted to conceal" where others build it to their own spec and vision. Its time for the vendors that add value to be paid for their work, and ensure that they stay in business. I shudder to think what would happen if we relied on Oracle to provide coherency in the developing world of open source software products.

Re:Oracle understands business

[turing_m]

Its time for the vendors that add value to be paid for their work, and ensure that they stay in business.

I agree. I think it would be better to CentOS and not have to go hat in hand to the executive board meetings than cause money to flow to Ellison just for slapping an Oracle logo on a product Redhat has developed. Of course, getting money to flow to Redhat would be a better outcome than either.

Re:Oracle understands business

[sgtrock]

Funny, I work for a U.S. company with more than 50,000 employees. We've got IBM mainframes, AIX, a little Solaris, a little HP-UX, Windows 2003 servers, and Red Hat Enterprise Linux. We've got some old databases on the mainframes that have been running the bread and butter apps of our business for more than 30 years, DB2, Oracle, MS-SQL, a little MySQL, a little postgresql. Our use of Oracle is limited to a single relatively large application that happens to be hosted on our HP-UX environment. Still, it's dwarfed by the traffic that flows through all of the other systems.

If I suggested to my senior management that we should move any of our OS support over to a database vendor who had just gotten into the game (and oh by the way, the OS is tweaked heavily to support a database that has very limited deployment), I'd be laughed out of the room. Especially when it's from a vendor like Oracle that has historically has given us rotten service in the first place.

We buy appliances for single purpose applications, not server farms. I'd be told to come back with a vendor with a proven track record.

So, let's revisit this question in 5 or 10 years. If Oracle has developed a proven track record of consistent support over time, developed a more generic distro (or created a black box deployment model), and vastly improved their customer service, we'll talk. :)

most U.S. business understands Red Hat == Linux

[rubycodez]

I've not seen Oracle Linux in any client of my company's data centers, and they include some with huge budgets (I.T. budgets over a billion). For running Oracle most are Red Hat, some are OpenSuSE, and a little bit of some others. No Oracle Linux anywhere.

One stop shopping

[Anonymous Coward]

If you run an Oracle shop with DB support, Oracle's Linux support is a deal that's hard to beat. It's comparatively cheap and coverage is 24/7 across all time zones until a problem is fixed no matter if it's database or OS related. Try that with other Linux support vendors. If it ends up being a DB issue, they'll point you to Oracle and tell you to have a nice day. Then you can start the trouble ticket process all over again and hope Oracle doesn't say it's an OS issue. Anyone up for finger pointing when your mission critical system is down?

Re:One stop shopping

[Anonymous Coward]

Well, last time I checked, Oracle didn't sell the underlying hardware or the network. There's two big items to point fingers at if they want to brush off your call.

If you're dead-set on top to bottom support, go DB2 on AIX on POWER hardware. Surprisingly (or not), you'll still be disappointed.

Re:One stop shopping

[Anonymous Coward]

or.. just hire a group of admins who understand what they are doing and use CentOS. Good admins arent cheap, but it's better than relying on someone else to fix your problems for you.

Re:One stop shopping

[nwetters]

"coverage is 24/7 across all time zones" Good luck with Oracle 24/7 support in the Middle East. In Qatar, you'll have to wait several days for a guy to get on a plane. Oracle's Linux support offers nothing over Radhat's in this corner of the world.

Re:cant we already get free and support with cento

[Anonymous Coward]

what about the btrfs filesystem? doesn't that do block level dedupe ? that's a very exciting feature IMO

They don't know what they are selling

[bol]

The greatest challenge that Red Hat (and Oracle) now face is to determine what they're actually selling and make a clear case for the added value that they provide.

I've run a few large Linux shops, recently including one requiring over 300 RHEL licenses and I can tell you that without a doubt that both Red Hat and Oracle sales people have zero idea what they are selling, what the differences may be and what added value they provide.

Red Hat copyrighted materials are the Red Hat trademark, logo, etc and the key difference between all of the RHEL derivates is simply the absence of that name and logo. Each distribution can pick and chose what patches and changes they want to merge in but everything is open source. It's how CentOS, Oracle, etc can make a competing "product." It's a bundle of freely available code and not much more. Where products differentiate is their delivery mechanisms and support of said code.

Things get complicated when you start asking Oracle and Red Hat what you're actually buying and what that support entails you to. I can tell you from first hand experience that I have never had a single issue get resolved via Red Hat's support organization - including clear bugs with tickets that still exist (primarily memory management code with kswapd.) Maybe they're only setup to help people get printers working with cups? And the same goes with Oracle Support.

By Oracle's move of choosing what code to merge and adopt they are misleading customers by openly calling it and comparing it to RHEL - which is exactly how it's sold and pitched to customers.

Oracle even offers a utility to run on your RHEL installation to re-brand it to Oracle Enterprise Linux. It replaces a bunch of packages and removes the Red Hat name, points it at the Oracle yum sources and calls it a day.

If Oracle wants to create a world class Linux they need to provide the tools, support and honesty to make it a successful competitor rather than relying on their name (which does not hold much clout, despite what their marketing guru's may think.) Combine that with resolution of real problems and not just entry-level technical support and you'll have a winner.

Come to think of it, that applies to Red hat as well.

Re:They don't know what they are selling

[eln]

I think you're misreading what Oracle is trying to do. Oracle is not particularly interested in creating the best Linux distribution out there. Oracle is interested in creating the best *end to end Enterprise solution* out there. Most of their acquisitions over the past several years have been toward that goal. Oracle wants to be the single source for every part of the software stack in Enterprise computing.

Right now, Oracle can offer a total end to end solution with one support contract for OS, DB, Middleware, and front end apps. No one else right now can do that, and that's a huge deal for the executives of the large companies that tend to run Oracle software. Oracle is not trying to compete with RedHat, Oracle is trying to compete with ERP providers like SAP. RedHat is just providing them with free OS development.

Re:They don't know what they are selling

[Anonymous Coward]

Right now, Oracle can offer a total end to end solution with one support contract for OS, DB, Middleware, and front end apps. No one else right now can do that

IBM?

Re:They don't know what they are selling

[Anonymous Coward]

No one else right now can do that,...

Except of course IBM - who who can supply and support everything under the sun:

• Hardware (System p, z10...)
• OS (AIX, z/OS)
• DB (DB2)
• Middleware (Websphere MQ/MsgBkr/ESB)
• Web-tier (Websphere Application Server)
• Custom application (via IBM Consultancy services)

even Microsoft has as much of the stack as Oracle with Windows, SQL Server, BizTalk and IIS.

Re:cant we already get free and support with cento

[baileydau]

i've never bothered to look at oracle linux, because i can get 'free' redhat through centos, and when i want paid support, i can get it directly through redhat.

without some other differentiation, what is oracle providing that isnt there from the others?

so yes, it is just a rip off of red hat.

You would ONLY use Oracle Linux to run your Oracle products on. You wouldn't use it for your file and print, or web server. They wouldn't want you to anyway.

It's largely a marketing thing. If you run your Oracle products on Oracle Linux, Oracle will support the entire software stack. That can be important to a lot of enterprise customers, no turf wars about who's fault it isn't.

As a bonus, the Oracle Linux support contract is (and should be) significantly cheaper than Red Hat (or Novell - the other supported Linux vendor). This is because they really only support those functions that are required to run the Oracle products. They aren't interested in supporting your file and print server etc. Whereas Red Hat and Novell have to support everything.

Can you imagine what Oracle would say if you had an issue that was borderline Oracle / OS and you were running CentOS? Even though CentOS is a re-badged Red Hat, it isn't Red Hat, and it isn't on Oracle's supported OS list.

The sensible thing to do would be to run Oracle Linux for your Oracle products and Red Hat (or CentOS if you didn't want support) for everything else. As they are all virtually the same, it's a lot easier for your administrators.

Oracle's kernel developers?

[tjstork]

Redhat employ _real_ kernel developers. Do oracle?

I would just throw something out there, but, Oracle pretty much is its own operating system in its own right. And, as such, it actually has to do concurrency, availability, all that ACID stuff that frankly "_real_ kernel developers" do not even bother with.

Yes, Oracle is a shitty company the U/I to this database is just terrible and always will be: but everyone knows that. We all have our Horracle stories. But, if you want to put a billion records into a database, and sleep at night, there's only one game in town, and that's Oracle. They've been doing MVCC now for almost 10 years, high availability, ROLAP stuff... been there, they did it.

Point is, if anyone knows anything about reliability, its going to be Oracle, more than it is Red Hat.

Re:Oracle's kernel developers?

[Anonymous Coward]

Actually, they are not the only game in town. There is DB2 on IBM mainframe. That really does provide peace of mind as long as you can pay the bills. The user interface is even worse and they are for people that think cobol is still a hip language, but IBM does understand what it means to deliver a fault tolerant system with hot standby spread over different sites. Oh and they are even longer in business than Oracle.

Re:Oracle's kernel developers?

[kestasjk]

But, if you want to put a billion records into a database, and sleep at night, there's only one game in town, and that's Oracle

The largest database I maintain for a site I coded has 15 billion records atm and it's doing fine in MySQL, with a relatively busy daily peak time with well over 100 users, all on shared hosting.

In fact the only problem I've had with database growth was when an auto incrementing ID went over ~2 billion in MySQL, which put it over PHP's 2^32-1 integer limit.
And yup this all has to do locking and transactions, not just MyISAM with basic queries.

My personal experience counts for nothing of course but Google, /., etc, etc all using free databases for big work too, so I think your attitude is a bit dated.

Re:Oracle's kernel developers?

[adrianmsmith]

"They've been doing MVCC now for almost 10 years"

Oracle introduced MVCC in Oracle 4, released 1984 (25 years ago)

Re:Oracle's kernel developers?

[Anonymous Coward]

Size is not all. Different queries make different system load. IBM and Oracle is the market leaders of large scale database systems. There are other players that also make large systems.

Large systems can be busy:
"UPSâ(TM) shipping system achieved a peak workload of 1.1 billion SQL statements per hour. UPS uses IBM DB2 Universal Database for z/OS database management software, IBM eServer z990 system and IBM storage."
http://www.wintercorp.com/PressReleases/ttp2005_pressrelease_091405.htm

Large systems can be large:
"The largest OLTP system is Land Registry for England and Wales. The 23.1 TB system employs IBM DB2 Universal Database for z/OS database management software, IBM eServer z990 system, and IBM and Hitachi storage."
http://www.wintercorp.com/PressReleases/ttp2005_pressrelease_091405.htm

Large systems can be expensive:
A example of a large scale IBM database server:
http://tpc.org/results/individual_results/IBM/IBM_595_20080610_ES.pdf
http://tpc.org/tpcc/results/tpcc_result_detail.asp?id=108061001

Re:Oracle's kernel developers?

[Macka]

IBM does understand what it means to deliver a fault tolerant system with hot standby spread over different sites.

Never heard of Oracle Data Guard then?

Re:cant we already get free and support with cento

[nprz]

Except that it is not ready for use beyond benchmark and review.
I tried it and like it. It is still missing removal of snapshots though. Without that, you will ultimately end up with a full disk.

Oracle != Linux

[John Utah]

Luckily, the only people I hear recommending Oracle's Linux is DBAs. All sys admins I know (including myself) wouldn't even consider it. RedHat all the way.

Re:Oracle != Linux

[pak9rabid]

All sys admins I know (including myself) wouldn't even consider it. RedHat all the way.

All the sysadmins I know (including myself) wouldn't ever consider it. Debian all the way.

Management on the other hand...

Re:Oracle != Linux

[arkane1234]

The thing is, not every application works on Debian. There are alot of applications out now that are binary-only and are purchased for Red Hat Enterprise.

Debian is great if you are managing applications in .deb format, or are either open source or available in source code in some way. Most business software has turned into requiring "industry standard" operating environments in order to be accepted by management. The reason for this is that it's easier to ask for Red Hat certification than it is for Debian certification, with the same results. That, along with the added bit of SOX compliance recently, and other hurdles thrown in that are adding in so many levels of troubles for management that it's easier to isolate them down to a given standard that is accepted by so many levels.
I'm an old school Linux admin (started with slackware in 1995-96) and I deal with Gentoo, Debian, and CentOS outside of work... at work, in highly visible positions I use CentOS or RHEL.

In not-so-visible situations, I use CentOS, Debian, and Gentoo.

Xen is a big deal

[btarval]

"Again, wrong. RHEL 5 ships with Xen, and will support Xen until at least 2014. OUL also ships with Xen. Please remember, KVM has not shipped in *any* RHEL release (major or minor) yet. Only Red Hat internally knows the release agenda."

I hate to correct an otherwise good post, but that is at best misleading, and at worst just plain wrong. Redhat has announced that they are only going to support existing Xen installations, while providing a way to migrate to KVM.

Xen is dead with Redhat. At least for now.

Personally, I think this is a major screwup by RH, as I know of sites which had been stongly RH but are now looking at dropping them. Sorry, KVM just isn't ready for serious primetime. What's worse, is that the majority of Virtualization research out there is centered around Xen, for the simple fact that it's been around longer.

So Xen is the focus of the next generation of technology, and will remain that way for a while.

And before the KVM fanatics jump up shouting the usual "but-it's-faster!" mantra, you should be aware that Type II hypervisor support (ala KVM) was announced a couple weeks ago at the Xen Summit (at Oracle's HQ, btw).

So one can either choose a KVM type of hypervisor, or the original Xen hypervisor.

Oh, and I heard that the guy who did it coded up in 12 days as a lark.

But unfortunately one doesn't seem to have a choice with Redhat..

I certainly hope CentOS picks up the Xen work from Fedora this year. Otherwise I'll have to look to Oracle for serious datacenter work. I'm not happy about that at all, as I've been a very strong fan of Redhat (and have given them lots of business.

But this really underscores how good it is sticking with Open Source. At least I DO have choices.

Re:Xen is a big deal

[Anonymous Coward]

> I hate to correct an otherwise good post, but
> that is at best misleading, and at worst just
> plain wrong. Redhat has announced that they are
> only going to support existing Xen installations,
> while providing a way to migrate to KVM.

I work for Red Hat; the information provided is not wrong. Xen within RHEL will be supported until, at least, 2014.

Re:Xen is a big deal

[btarval]

Either RedHat's Marketing department is seriously misleading, or you're seriously mistaken.

. Again, the question isn't whether old existing installations will be supported. It's about RedHat dropping Xen for future installations.

Let me give you this Marketing blurb, since you don't seem to be aware of it: Red Hat Sets Its Virtualization Agenda "Red Hat's strategic direction for the future development of its virtualization product portfolio is based on KVM,

and

"Existing Xen-based deployments will continue to be supported for the full lifetime of Red Hat Enterprise Linux 5, and Red Hat will provide a variety of tools and services to enable customers to migrate from their Red Hat Enterprise Linux 5 Xen deployment to KVM."

Note the words "existing" and RHEL5. No mention about the future except for migrating to KVM.

It's pretty clear that, going forward, if you want to stay with RedHat, you need to move away from Xen.

Sorry, but many people are going to stay with Xen over RedHat. Fortunately there are companies who are willing to accomodate them.

If you have some official news to the contrary, I'd appreciate hearing about. Because right now, people are looking at (and finding) alternatives to Redhat.

Re:Xen is a big deal

[ak4good]

Xen is not going anywhere because of the size of its install base, if no other reason. RedHat is just hurting itself by dropping it. Go Debian.

Re:Xen is a big deal

[Macka]

Given that the person from Redhat said that Xen will be supported in RHEL till 2014, and the information you posted says that Xen will be supported for the lifetime of RHEL 5: one can draw the conclusion that the person from Redhat is right because RHEL 5.n will be supported for that long.

KVM is great and all, but I'm sure Redhat don't expect customers to dump Xen until they've had a decent return on their investment. Redhat are not dumb.

Re:Xen is a big deal

[Anonymous Coward]

You miss the point. Which is that RH is not claiming that Xen will be supported beyond RH5.

Can you show me one statement where it's going to be supported in RH6? That seems to be lacking. And that's why they've limited their statements to existing installations and not new ones.

RH used to be smart, but there's a different crowd running it now. Haven't you noticed the key engineers who have left?

In short, they've bought up the company which developed KVM, and are trying to get their customer base to use that instead, in order to recoup their investment.

It's just not going to work, and it will cost them customers who need a commitment beyond RHEL 5.

Re:Xen is a big deal

[StillAnonymous]

What's the big deal? RHEL6 will be KVM based, but there will be a way to migrate your VMs from Xen over to it (one of the really nice features of virtualization is that moving virtual disks around is so easy).

KVM is part of the default Linux kernel now, and Xen isn't, so there's none of this hassle with backporting Xen patches to older kernels and having to do extra testing with each release.

In the end, most of this should be pretty transparent.

Re:Xen is a big deal

[btarval]

Because there's a lot of interesting research and new technology coming out which is Xen based and not KVM. It looks like things will remain that way for some time.

Or, to look at it another way, KVM is behind the tech curve, and has a ways to go to catch up.

Then there's the issue of having to change over to a different infrastructure. Do you have any idea of what it takes to handle the vast number of VMs in Cloud Computing?

All in all, it's just better to go with a vendor which is more committed to your requirements.

Re:Xen is a big deal

[Macka]

You miss the point. Which is that RH is not claiming that Xen will be supported beyond RH5. Can you show me one statement where it's going to be supported in RH6?

Um, no I'm not missing the point. I know they're not claiming it's going to be supported in RH6, which I understand to mean that it won't be. But it will be supported in RH 5.{7,8,9} or whatever the release is in 5 years time. And 5 years is an age in this business.

It's just not going to work, and it will cost them customers who need a commitment beyond RHEL5

I don't think so. Businesses implementing new builds today will be looking for a technology refresh 5 years from now, and by then I fully expect Redhat to have proven tools at their disposal that can import/convert Xen VMs into KVM VMs in the same way that VMware can import Parallels VMs now. There will be a migration path.

Re:Xen is a big deal

[Macka]

Or, to look at it another way, KVM is behind the tech curve, and has a ways to go to catch up.

I used to work on Tru64 Unix on Alpha, back in the day when it was (arguably) the best platform amongst its peers. Certainly the Alpha chip was. What I learned from that experience is that where business and money is involved, the best technical solutions are not always the winners. That said, the fact that all of KVM is part of the core Linux kernel now means that it will benefit from constant and continuous development and improvement. So I fully anticipate that it will eclipse Xen at some point in the relatively near future, which is why I personally am paying a lot more attention to it than I have in recent years.

All in all, it's just better to go with a vendor which is more committed to your requirements

Agreed, and as SuSE Novel have thrown their weight behind Xen, that would appear to be the distro of choice for you.

Re:Xen is a big deal

[Macka]

KVM is part of the default Linux kernel now, and Xen isn't

Um, that's not strictly true. The domU Xen Code is already in the mainline kernel, and there's been some recent discussion to include dom0 as well. Based on a recent LWN article ( Xen: finishing the job ) I'd say there's a very good chance that will happen.

Re:Xen is a big deal

[Anonymous Coward]

I agree with your intended meaning, but you ended up suggesting basing a major enterprise deployment on software developed 'in 12 days as a lark' as a credible option.

Re:cant we already get free and support with cento

[Anonymous Coward]

Oracle Cluster Fuck System? No thanks! I'll keep on using the non-cluster-fuck version of oracle. At least then it's only one fucking.

where to start...

[msimm]

If you don't need them, avoid them like a money grubbing plague. And pay someone who really provides value to the Linux (our) community.

Problem with Oracle EL

[mysidia]

Last I checked, it didn't come with a Yum setup, and you had to pay for support to get simple things like software updates and the ability to install software from their FTP servers, or from anything other their install media.

$99 per year per system for an update-only contract, and without it, you can't upgrade to apply things like security patches (until they eventually release a new version of the whole system with new media, for you to update using install media). That's not free!

With CentOS, Scientific Linux, etc, you actually get your security fixes automatically with yum update. With Ubuntu and Debian you get 'apt-get'.

Until/unless Oracle does something about this deficiency, there's really no reason to pick Oracle over the alternatives, except in the event you have to pay for support anyways.

I am. No problems (so far).

[emil]

I had Red Hat on a 1u dual xeon manufactured by IBM. Minimal load, but the box would crash every 6-9 months. I never bothered to figure out why; just rebooted.

While I was migrating to a dual socket, quad core (also by IBM), my subscription died. I learned that someone at corporate HQ had terminated my RHN up2date license (among many others). I admit that I did try to get Red Hat support turned back on, but I couldn't even get their sales staff to send me a quote by my deadline.

Oracle, however, was quite timely and I was able to throw a few RPMS on my new RedHat install and get it patched. The licensing is nebulous, but the system has remained reliable. I have higher tolerance for a bit of downtime, however; ymmv.

Re:I am. No problems (so far).

[Anonymous Coward]

You're not alone in your experience with Red Hat sales.

Re:cant we already get free and support with cento

[doktorjayd]

The sensible thing to do would be to run Oracle Linux for your Oracle products and Red Hat (or CentOS if you didn't want support) for everything else. As they are all virtually the same, it's a lot easier for your administrators.

IMHO, the really sensible thing to do is not run oracle products at all. even the bea purchase and rebadging of the weblogic/aqualogic app server doesnt change that.

Re:cant we already get free and support with cento

[nicolas.kassis]

But CentOS is in no way affiliated with Red Hat. All they share is the code. CentOS is fine if you want to go the free route but redhat is pretty cheap. We order Red Hat licenses by the dozen for our University. All Oracle servers here run on Sun because of legacy I guess.

Oracle wants everything...

[Anonymous Coward]

They make their own version of Linux and their own virtual machine based on Xen. They won't support Oracle on any other virtual machine other than their own either...assholes.

Why would I bother with Oracle's version of Linux for the database and then a different one for web servers, app servers, etc.? Just use Redhat or even better Fedora or Centos for free. Why pay for Redhat at all? I guess there is a sucker born every minute but in 20 years of IT I have never called a vendor for OS support, maybe I've just been lucky.

Support for software is only worth the upgrades anyway. Everytime you call any software vendor they will just ask if you have the latest version installed. If you do, then the fix is always in the next version. If not, they will tell you to upgrade to the latest version. That is software support.

Re:cant we already get free and support with cento

[MoreDruid]

That's the point: they are definately NOT the same. Yes, its based on Red Hat, but there are so many differences on Unbreakable Linux (here we call it Broken Linux) that annoy the hell out of you. The cluster application Oracle sold to our customer is not cluster aware (how did they do that?), furthermore the cluster service needs to be restarted when a node goes down (WTF? what's the point in having a cluster if one of your nodes can't fail???) and there are more diffences and issues. I've worked with it over half a year now. I've only had issues for those past 8 months with those Unbreakable Linux systems. In the meantime a cluster 3x larger - and much heavier used - is running without a hitch on a Red Hat base install and the same whole Oracle shebang on top of it. Go figure.

Re:cant we already get free and support with cento

[reiisi]

Right. If you can't afford support you don't need and you can't afford the time Fedora takes, CentOS is great.

Ordering red hat licenses is one way to make sure the OS is still there for you next year. If you're using it in business and making much profit (or just saving money) by using their data products, you should be recognizing that you need to give them (or canonical or one of the others) money because you need them to be there next year.

Same with feeding bugs back by using Fedora. If you rely on the OS as a tool in your job, you want to help keep the project alive and healthy.

So, actually, even if you are using CentOS, your self-interest will induce you to support the community in whatever ways you can afford to, maybe even just by helping others start using open source.

(And while it would actually make sense for Oracle to have their own distribution based on Red Hat, it does not make sense for them to be effectively dissing Red Hat. Unless, I suppose, their share-holders and/or primary customers expect Ellison to put on the dog.)

Re:Sad :(

[reiisi]

My company has been moving from Solaris to Oracle Linux recently.

Sad :(

grep wtfismystorage /proc/scsi/scsi; nohup stabselfinface &

<nods-head/>

Red Hat Sales crew

[reiisi]

Wondering if Red Hat's sales department needs more people who understand how to sell "free" software.

I mean, maybe there are two problems here:

One, maybe they are too short-handed to meet demand.

And, two, maybe they are short-handed because many people who understand the benefits of free software would rather be using it than selling it.

Cent OS slow?

[reiisi]

Hmm. Maybe they should be charging more so that they can afford the manpower to move the updates and patches down-stream?

Or maybe their users should be donating more?

To make things a little more clear,

[reiisi]

Oracle is the source of the controversy. They are the ones strutting around saying, look what we're getting for free!

I suppose it is because many of their big customers expect them to play the predator. It's not the money saved. That's peanuts.

It's the image. Oracle provides a buffer between the dog-eat-dog corporate world and the touchy-feely alternate corporate world.

Re:cant we already get free and support with cento

[Splab]

So what enterprise class DB would you run instead?

Uhm, no. Don't evangelize Oracle Linux.

[reiisi]

Oracle's Linux is tuned for Oracle's primary product. More than half of your servers do not run Oracle's database. Saying, "Let's run Oracle Linux!" for anything but Oracle's database servers could be bordering on incompetence.

And if your management is still saying "What's that?" when someone suggests implementing servers on Red Hat's OS, maybe you need to communicate more positively with management about the state of the OS marketplace.

Just saying.

Re:Uhm, no. Don't evangelize Oracle Linux.

[See Attached]

Why would Oracle chose to tune Redhat OS and Linux kernel in secret? Changing things like the "ld" binary seem a bit underhanded at best. Oracle is undermining the service provided by Redhat. They lend it credibility by copying it/ hacking it, and then fiddle with its revenue stream, as if they deserve it? Its obvious they add little to the big picture, and have a track record for producing products that are overly complex. IMHO.

A couple of reasons we will not migrate *ever*

[wzzrd]

1. If you have a mixed server park (that is: host different applications too, other than Oracle's), migrating is *not* feasible. I'm not going to support yet another OS, just because it is *possibly* a tiny bit more convenient once it is set up. Because before it's set up, I'll need to have deployment mechanisms for another OS, management tools for another OS etc. Not worth it.
2. To extend my first point: Oracle's support might be a bit cheaper for the OS, my time is a lot more expensive than a thousand bucks worth of support on a years basis. That matters when having to support more OS'es.
3. Red Hat fixes the bugs, and then releases the src.rpm. Oracle has to Q&A that, port it, upload and release it. Updates for Oracle Linux will be (a lot) later than Red Hat's. See how much time it is costing CentOS to release 5u3. No offence, but for production systems, I want to have potential fixes *now* if the situation we're in is hurting us.
4. I'm just about to get RHCA certified. Can I get that level of Linux certification from Oracle? Don't start saying the OS'es are compatible, because they are not, see point 5 and 6.
5. The only thing Oracle can do on the long term, if fork RHEL. The amount of support, the changes they make and the fact they want to support until the end of time in the own way, might not be called a fork, but it will be just that in the end. So much for compatibility.
6. They ported yast. Need I say more?
7. Not really a business reason but check this out. Oracle announced Oracle Linux just a couple of months after RH scooped JBoss from underneath Larry's nose. One of the previous posts is right: Oracle is not trying to compete with RH. It's trying to get revenge. ;-)

Re:A couple of reasons we will not migrate *ever*

[BitZtream]

3. Red Hat fixes the bugs, and then releases the src.rpm. Oracle has to Q&A that, port it, upload and release it. Updates for Oracle Linux will be (a lot) later than Red Hat's. See how much time it is costing CentOS to release 5u3. No offence, but for production systems, I want to have potential fixes *now* if the situation we're in is hurting us.

You get the potential fix at the same time redhat does. Its GPL remember? they can't release it to anyone unless its everyone, so you indeed get the same patchs as the redhat people at the same time even if you are running Oracles flavor.

You are completely free to apply them as well, again you have the source. If it works, great, if it introduces bugs, well that sucks.

Or, you can wait till Oracle releases a patch that has been QA'd as you said and have less problems on your Oracle servers.

You're point in #3 is just wrong, if not a guideline for how to be a shitty admin.

As for point #4, you do realize that these certifications are ways for redhat to make money right? They don't actually mean you have a clue, as per your previous statement.

#5 They don't have to fork, theres no real that says 'If the glove fits you must fork!'. They can just keep doing what they do and tracking Redhat. Redhat is more than likely going to include any fixes Oracle comes up with because Redhat wants Oracle to run well on their machines as well.

#6 As a FreeBSD user, I could retort with 'Yes, but the morons are running everything on top of Linux so its crap regardless'. But must like your statement that would just be a retarded fanboy statement with no actual value to the post.

#7 Take your tinfoil hat off, you aren't that clever, it was pretty obvious to everyone else but you I guess.

Right. IBM.

[reiisi]

Oracle is probably trying to position itself against IBM.

But they really should be passing some of their revenue stream back upstream, instead of pretending to be predatory towards Red Hat.

Maybe they plan to eventually make a play for Red Hat. That would be a shame, because, unless Oracle changes a lot internally, they would not know what to do with Red Hat.

Even IBM knows better than to try that.

Re:Right. IBM.

[alexborges]

Oh theyve tried, believe us, to take over redhat many times.

Theyve been shown to the door on all those ocassions. And they will still be shown out this time, now by customers.

GOOD.

I mean, their offer on this is so idiotic that its scary: where the hell will they get the manpower to actually give YOU OS support? How does saving 2500 bucks/year (what rhels most expensive intel suscription costs), versus a HUGE support plan for the oracle DB (that is more into de tens of thousends per-year/CPU), while reducing the posibility to get the best possible platform support, makes sense to the customer?

Youd have to be a suckpuppet for the brand to actually buy into this.

(well, to not attribute to evilness what can be explained by stupidity: you may just as well be plain stupid)

Re:cant we already get free and support with cento

[turing_m]

So what enterprise class DB would you run instead?

PostgreSQL, unless there is some feature PostgreSQL is missing that I would need for the given application in the foreseeable future.

I sense a disturbance of the force

[alexborges]

IBM brings Sun back from the dead, Oracle tries (yet again), to surface as some company that can actually support an OS. Novell is still being breastfed by the big names in the industry, despite it delivering an unsufferable piece of shit of an OS, consistently, for years now.

RH, on the other hand, got its hand on QMranet (whatever), and has the SolidICE thingie that, from my POV, will bolster it forward as a virtualization leader that will break the potential lockin embedded in closed-source VM image disks, formats and proprietary programs and the like.

The powers-that-be are scared, and want to slow Red Hat down.... i dont think its gonna work: the numbers on RHEL this year will be sky-high, both on profits and on addoption: its redhat who will capitalize better in this times of crisis, and oracle, ibm, microsoft and the like, who stand to loose the most in the corporate space.

Thanks MS and associated Repugs, you first gave us Vista, then you gave us this huge crisis that will force the market to actually look at the value you deliver... scary thought, yes?

Well, you SHOULD BE SCARED.

The grep

[Kludge]

] grep -ri red linux-2.6.28 | grep -i hat | wc -l
3877
] grep -ri oracle linux-2.6.28 | wc -l
191

Re:cant we already get free and support with cento

[Anonymous Coward]

Just FYI--- "Unbreakable Linux" is the support service. "Oracle Enterprise Linux" is the distribution.

The cluster application Oracle sold to our customer is not cluster aware
What cluster application?

What about support for all LSB compliant distros?

[TTimo]

We use regular RHEL to run our Oracle database. When we were setting this up, several people pointed out that Oracle's Linux was doing some pretty horrible things to the kernel and overall system setup, so we stayed away from it.

As far as support, lets face it: Oracle doesn't provide support. You can open tickets with them, and maybe get pointed to a patch to resolve a problem. If you are not a database shop you will be working with an independent vendor, who will setup the DB and do a lot of the administration stuff for you. We're pretty happy with the guys we use, and they didn't care either way between Oracle's release or "true" RHEL.

The larger issue is, why not support ANY LSB compliant distro? I dislike RHEL for various reasons, I would much rather run Debian like I do everywhere else. However I can't because of the support contracts. It wouldn't make any difference to our DB support people or Oracle operation, but I wouldn't waste my time figuring out a different distro each time I have to do something on the DB nodes.

Their product is just not that good

[Douglas Goodall]

I was an early adopter of the unbreakable linux. I had serious support issues just trying to get my Dell SC1420 to use it's SATA drives in my desired configuration, and despite my service contract, was unable to get either support for that, or reliable updates from oracle and their bastardised version of up2date never worked for me. I had high hopes for their version, but they were all talk and no action when it came to keeping up with the upstream provider. When their system boots, it still identifies itself as Red Hat and they never even properly customized the system enough to cover up that fact. They (Oracle) just used the open source situation to grab Red Hat's product and try to run away with it. I dislike Oracle and will never use their products again.

Re:Their product is just not that good

[HAWAT.THUFIR]

When their system boots, it still identifies itself as Red Hat and they never even properly customized the system enough to cover up that fact. They (Oracle) just used the open source situation to grab Red Hat's product and try to run away with it. I dislike Oracle and will never use their products again.

Sounds like a lawsuit waiting to happen, there. Basically, they have to repeat the efforts of whitebox linux. Maybe they could form some sort of partnership with whitebox so that they don't have to duplicate that effort? Seems like it would be easier to cut a deal with Redhat. -Thufir

It sure could have been a smart move.

[reiisi]

Conceptually, it is a good idea to have their own distribution.

Conceptually, Red Hat would be a good choice of a distribution to fork.

Conceptually, it's a good idea.

But, practically speaking, they're screwing it all up.

No, it is not good business to take without giving.

No, it doesn't reduce the price of Oracle's server stack significantly to cut Red Hat out.

Not significantly, not with what they lose by cutting Red Hat out.

No, it isn't business savvy. It's cutting off their nose to spite their face. And it's totally misunderstanding the meaning of free as in freedom, not as in beer.

In essence, it's trying to give their customers free beer and put it on Red Hat's tab.

Now, maybe they hope to absorb enough of Red Hat's business to induce Red Hat to sell the company. But that kind of predatory business always comes back around to bite you in the end, and it eventually destroys your own business.

Maybe they are actually feeding some of the revenue back to Red Hat. If that's the case, I'll take back part of this rant. But they still have said a lot of things publicly that sound more like a spoiled rich kid celebrating that he gets to legally take the poor kid's candy. And the AC who said it was business savvy seems to be doing the same thing.

Re:It sure could have been a smart move.

[BitZtream]

You've just described for GPL the exact same thing GPL supporters say about the BSD license.

Guess you should have just used the BSD license to begin with and left the GPL virus off of it, would have accomplished the same thing wouldn't it?

Except, that what you say is entirely untrue. Oracle makes a fix for something in the Redhat base, guess what? Redhat can take it and use it! Its gonna be GPL'd too remember? So Oracle isn't feeding off Redhat anymore than Redhat feeds for free from Linus himself (and all the other contributors). This is exactly what GPL promotes and the viral nature of GPL actually protects Redhat (and all of the Linux domain) in this case as it requires that Oracle make those changes available under GPL, which means Redhat didn't have to support the customer directly, but still gets the benifits from Oracle doing so in the form of a patch.

Note: GPL is an annoying virus to me. It restricts freedom to reuse code just as much as most other commercial licenses and is just as much a pain in the ass to the business world. However, what Oracle is doing is EXACTLY the freedom that GPL tries to promote. Seems rather stupid to bitch about the 'abuse' of the license which results in an almost perfect real world example of how it was intended to be used.

Hmmmm, where have we heard that before?

[Nick Driver]

If you can have one provider who will offer support for the entire stack, OS, virtualization, database or middleware engine, you have a huge win on your hands.

Microsoft certainly likes the way you think.

I recently built an Oracle-based system on AMD 64-bit HP Proliant server hardware, using Oracle's Enterprise Linux and the 11g database and it's been running strong and fast for three months without a single reboot yet, and is hosting an entire accounting/purchasing/finance/payroll system for a city govt in Texas.

The Oracle "stack" definitely works extremely well, and this particular installation replaces an old AIX-based platform where the hardware alone cost nearly $160K when it was purchased 8 years ago. The new commodity-grade hardware for running Linux cost less than $25K for quadruple the storage capacity and over triple the performance.

Cluster filesystem clusterf*ck

[mcrbids]

Clustered filesystems are, as a breed, ridiculously over complicated. perhaps king of the hill is OCFS. To get it working right, your entire cluster has to to perform a series of steps IN SYNCH. EG: your entire cluster must all be done with step 1 before they all do step 2, etc. Just too complex, and no way to be redundant without blowing loads of cash on highly complex hardware....

Sorry... NO!

If you want simple, redundant storage, you really have to do it in the application layer. Doing it at the OS level requires too much abstraction to do well while maintaining decent performance. The closest I've seen to a decent clustered filesystem from an administrative standpoint is Gluster F/S...

Re:Cluster filesystem clusterf*ck

[alexborges]

And anyhow, Red Hat GFS is MUCH, MUCH, MUCH better (even faster), than OCFS.

Re:Cluster filesystem clusterf*ck

[jabuzz]

And IBM's GPFS pisses over both GFS and OCFS from orbit.

Re:Cluster filesystem clusterf*ck

[mcrbids]

GLUSTER FS IS NOT GFS.

Please read up on it. GFS is better than OCFS, but Gluster FS is WAY better than both.

talking to oracle

[nimbius]

last year regarding their OS led me to believe they have absolutely no technical expertise in linux. we did not have a technical expert present at the discussion, only 'senior support' personnel.

the conversation was always steered toward response times and support metrics...extensive testing and development. nagging kernel memory use questions never got responded to. BTW, youre getting a linux with no SELinux, so essentially oracle is shipping a "preconfigured" hacked red hat to eliminate the confusion most developers have when they realize their windows oracle expertise stopped at "insert CD rom" in the linux realm.

oracle in linux performs no better than oracle in windows...its load handling sucks and the uptime for it was often measured in hours. maybe a bad developer or two to blame, but Oracle linux did not help. dropping SELinux got us dinged on SOX compliance as well.

Re:talking to oracle

[Anonymous Coward]

Sounds like you need support more then most as your post shows that you know very little about Oracle sales, its Linux distribution or database. If you were "talking" to Oracle support sales, response time, metrics and price would be appropriate topics. Did you really expect a sales guy to analyse your system, fix your memory problem and tune your database?

Oh and by the way, SELinux is fully implemented and on by default in the Oracle distro just as it is with RH, so it's obvious you've never actually installed it:

http://selinuxnews.org/wp/index.php/2008/11/17/oracle-enteprise-51-linux-certified-at-eal4/

From all the FUD, I'd bet your poor uptimes are due lack of experience and skill on your part rather then any issues with Oracle database that somehow manages to scale and run reliably in the largest of enterprise environments.

Re:talking to oracle

[Anonymous Coward]

oracle in linux performs no better than oracle in windows...its load handling sucks and the uptime for it was often measured in hours.

I call bullshit on this. Our production environment is 75% Oracle under Red Hat, and our uptime is just fine, thanks. The only issues we've had were due to defective hardware and developer fuckups (ours, not Oracle's or Red Hat's).

Reliability is at least equal to the other 25%, which is HPUX.

Incidentally, we are transitioning from Red Hat to Oracle Linux. Although it sounds good in principle (having one vendor responsible for the entire software stack), I have serious reservations about Oracle's ability to support kernel issues, particularly after experiencing their RDBMS support. Unfortunately, those who made the decision aren't particularly interested in my opinion.

So for three years . . .

[Rambo Tribble]

. . . they've kept this revelation a secret while they attacked Red Hat on more superficial terms? How convenient. Methinks the Oracloid doth protest too much.

Oracle 'Support' is worthless

[Anonymous Coward]

Oracle Support is terrible. Stay away.

Re:Oracle 'Support' is worthless

[Octopuz]

That's not a fact but an opinion. Please try to give factual information when posting plain opinions. Okay, it may have become worse since they moved parts to Romania and Egypt. All good engineers that left in e.g. other European countries were replaced by Egyptians who were much cheaper but unfortunately less skilled, both technical and in communication with the customer.

Re:Oracle 'Support' is worthless

[Anonymous Coward]

nah, in this case it's fact.

Re:cant we already get free and support with cento

[Anonymous Coward]

Did you know that Oracle does printing?

This is because they really only support those functions that are required to run the Oracle products. They aren't interested in supporting your file and print server etc. Whereas Red Hat and Novell have to support everything.

Have you ever tried to open a support ticket with Oracle on the printing support in Oracle?

Usually the tickets got closed faster than WORKSFORME tags gets slapped on poorly maintained F/OSS application bug reports. The best support reply from Oracle I ever received on Oracle printing was "it's in the product, but you will have to support it yourself."

The sensible thing to do would be to run Oracle Linux for your Oracle products and Red Hat (or CentOS if you didn't want support) for everything else. As they are all virtually the same, it's a lot easier for your administrators.

Support calls come from when your administrators can't solve the problem. Even on the Oracle Linux systems people still install a third party printing system, just to get the functionality for which your company is paying so much. After all, third party support from someone who can solve the problem is better than paying a lot of no real support at all, right? Oh, wait, this is Oracle we're talking about.

Re:cant we already get free and support with cento

[doktorjayd]

define 'enterprise class'.

without the markitecture diagrams and execuspeak please.

the vast ( and i mean _vast_! ) majority of needs in an application database are covered just fine by free/open source tools, speed, reliability and redundancy inclusive.

the remainder probably could benefit from reconsideration of architecture and data access design.

the only reason you would nominate oracle is if your department needed to blow a remaining budget in order to get the same budget next year.

that or oracle sales reps helped write your requirements documentation.

'enterprise class' pffft.

Re:cant we already get free and support with cento

[Splab]

PgSQL does not scale out, only up. Enterprise is out of the question here.

Re:cant we already get free and support with cento

[Splab]

So what open source database supports those things?

MySQL has speed and redundancy (but not subsecond fail over and it needs some tool to handle the failover for it), but MySQL lacks reliability.

PgSQL has an ok speed, it is very reliable, but it lacks redundancy (yes there are people who have written all sorts of trigger based replication schemes for PgSQL, but they are unsafe, and like the MySQL option needs something else to handle the switch to primary).

There might be some obscure OSS database that handles these things, but I haven't heard of them yet.

Enterprise / Carrier grade databases are databases that even during catastrophic failures (fire in one data center, massive hardware failure in one server etc) are able to keep turning out transactions - at slower rates, but transactions are still going.

implicit obligations

[reiisi]

BSD or GPL, it makes no difference. The GPL makes the obligation explicit, but a BSD class license does not remove the consequence of failing to give back. In the end, you lose if you fail to give back.

No, Red Hat can't take Oracle's patches, most of them, because they are completely useless to Red Hat. The patches are specifically to support Oracle, and do not generalize.

How do you fail to see what Red Hat gives back to the community?

When you call the GPL a virus, you appear admit that you prefer to be able to ignore the obligation, and the effect of ignoring on the future of the tools you use.

I'd suggest you re-think what you choose to be annoyed by. The BSD class licenses do not buy you any better future than the GPL licenses, even if they may give you a little more leeway in the present.

Re:implicit obligations

[PCM2]

No, Red Hat can't take Oracle's patches, most of them, because they are completely useless to Red Hat. The patches are specifically to support Oracle, and do not generalize.

Wait. So no Red Hat customers use Oracle? No Red Hat customers could benefit from Oracle's patches? Then it sounds like all of Oracle's efforts are a complete waste of time -- and still, Red Hat loses nothing.

Define "Enterprise"

[turing_m]

Enterprise is out of the question here.

"Enterprise Class" is marketing gibberish designed to produce FUD. There are many, many databases used in a typical "Enterprise" that do not need to scale out and never will. In fact, I would bet more companies have need of this sort of database than the other. Every company has need of databases that cover internal control issues - accounting, inventory, payroll, industry specific stuff. These do not need replication. (Often they just need a better backend than MS Access.)

The primary concern of these databases is data integrity - that is, that the RDBMS does not treat your data like junk, and that you can constrain your data however you require. PostgreSQL has a well deserved reputation there. The secondary requirement is flexibility - can you get at your data and do what you want with it to learn more about your business or save you time? With flexibility in querying and a wide range of languages for stored procedures, PostgreSQL has an excellent reputation here too. Speedwise it is also no slouch.

Besides, if you really need to scale out (now or in future) there are third party solutions such as Slony, and now the PostgreSQL team is working on simple, built-in replication.

What are you arguing?

[reiisi]

What percentage of Red Hat installations are running Oracle?

What percentage of those are running OCFS, as opposed to GFS?

And so forth.

So far, they've been lucky with their patches, and haven't been too far behind the bugs the customer sees. But lucky is not proper support. Putting their own support crew on the phones is great, and showing their confidence in (the enterprise market leader) Red Hat is nice in some ways, but there is a sense that they should do more.

Maybe it really isn't necessary, maybe the patches are al that are necessary, but it seems to me that they could doing a lot more work (and releasing that work to the community) if they built their own distribution.

The differentiation gained from the fork should make their contributions much more accessible. As it is, everybody has to look at the patches and scratch their heads and do some real research to figure out how to use them. (Or do you have the impression that the community should just trust Oracle and apply all the patches?)