I'm pleased to see that evil deeds have consequences, though I'd like to see Oracle suffer a little (OK, a lot) more before getting over the hump. People don't use Oracle Linux because the Open Source community doesn't trust Oracle (and they shouldn't; that mistrust has been well-earned over a couple of decades).
There's a lot of reasons not to use Oracle Linux, most of them non-technical. Oracle is simply not an ethical member of the Open Source community, and if you trust them, they will screw you, some day, some way. Red Hat may have their flaws, but they've never sued over patents and they've never attempted to destroy competing projects or companies through legal threats and bullying. CentOS may be slow to jump on updates sometimes and to get out new releases, but at least it's a good Open Source citizen.
As an aside, if you want a faster moving RHEL rebuild that has paid developers working on it, you might try Scientific Linux. It is built by CERN and Fermilab, and tends to be very solid and fast to update and invisible (i.e., I don't think about it, at all, and it Just Works). We switched from CentOS to SL back when CentOS 6 was so late being released; couldn't be happier with it. We added support in Virtualmin for SL for just that reason...so many of our customers wanted 6, it was worth the effort to add a new OS.
So, yeah, Oracle is gonna have to have a "come to Jesus" moment if they want to participate in the Open Source community. They've got a lot of repenting to do.
This is the exact same feeling that I'm getting from this announcement. To me this just feels like a publicity stunt, where Oracle is aiming to get a slice of the CentOS/RHEL market with very little investment.
As far as I'm concerned this isn't even a valid attempt on their behalf to earn the trust of the Open Source Community. Taking something that already works and touting your own horn about how you can provide faster updates isn't really something that a big corporate player that wants to get into the open source market should be doing. I'm sure there are plenty of technologies at Oracle, that if open-sourced would have a much bigger impact, and would put them in a better relationship with the open source community.
Finally, when you are an enterprise/corporate player, and your press release for a new product is littered with "this is not a gimmick" there is definitely something wrong. In all honesty this just sounds like one of these "come with me kid I'll give you some candy! It's gonna be ok don't worry, just a little further". This is not something I would expect from such a big company.
I was trying to go for a "for linux nerds, by linux nerds" vibe, so I'm amused that your complaint is "This doesn't sound corporate enough".
At the end of the day, my view here is that Oracle has actually produced something useful, but lots of people are blind to it, in no small part because of what ultimately boils down to zealotry (which doesn't seem like an awesome reason to me). Hence the point of trying to get this out there.
I assume you're on the Oracle Linux team? I'm sure you're sincere, and I'm sure the Oracle Linux team has many good folks working on it (likewise, the rest of Oracle). But, when you work within the belly of a beast, an absolute horror show of historic and ongoing wrongdoing, you have to expect pushback and mistrust from the Open Source community.
Oracle is a corporate sociopath. If you wish to label me a zealot for expecting ethics from the people and companies I work with, that's fine. But, it's not going to alter the reality that I am not alone. Many people mistrust Oracle, and just because one unit within Oracle seems to be trying to do right, it doesn't alter the fundamental nature of that creature. It'd be easier to overlook past misdeeds if Oracle was not currently behaving in unethical ways on a massive scale, and attacking Open Source on several fronts.
Oracle cannot have it both ways. It cannot wage war against Open Source and software freedom, and expect the Open Source community to just look the other way and choose Oracle products. At least, I sure as hell won't be looking the other way.
Look, with all due respect we're not blind to it. I'm sure Oracle Linux is a fine piece of software, I bet it's just as good as RHEL and I'll even spot you that your support team is equally good as Red Hat's. We do not have technical objections to the work you've done.
We simply don't trust your employer to behave with anything even approaching good faith in any interaction. What happened to Solaris alone is a huge warning flag, those of us who have interacted with Oracle in a professional capacity have...more reasons to believe this.
No, not FUD...not really. Solaris is a complicated tale.
The short explanation is that when Oracle acquired Sun, all of the hopes and dreams of Solaris users, particularly OpenSolaris users, were dashed. Many great technologies were pushed onto the back burner. All of Sun's Open Source portfolio saw a massive shake up and very little came out the other side healthier. Pre-acquisition, Solaris/OpenSolaris was a reasonable OS choice for new server deployments, particularly in cloud environments. Post-acquisition, you'd have to be a little bit nuts to deploy Solaris. Solaris has a lot of die-hard fans, with good reason...so this was a bitter pill to swallow for a lot of folks.
Oracle is simply ham-fisted when it comes to dealing with Open Source technologies that it comes into possession of. It's a really good example of a company that doesn't "get" Open Source. It's not limited to Solaris, Solaris is just one of the most popular, and the most difficult to replace for people who are relying on it.
I am a long-term SunOS/Solaris user (~20 years) and what killed Solaris is they neglected Solx86 for years for fear of cannibalizing SPARC sales. What happened was devs got Linux boxes to replace SPARCstations (using them as cheap X servers) then x86 and x64 got good enough for servers, and everyone ditched Solaris in order to get onto the commodity hardware (Linux wasn't really a factor in this decision, it just happened to be the dominant Unix on that hardware at the time, could easily have been FreeBSD). Thanks to Microsoft and their NT pretensions, supporting hardware vendors (e.g. storage) were manufacturing compatible kit too. It didn't help that Sun managed to forget that they were a business, and a hardware business at that. Giving stuff away to drive sales is a proven strategy, but no-one was buying SPARCs. Turns out it was cheaper to buy PCs and accept failures and workaround them, than to buy a "real" server that wouldn't fail and if it did, you could hot-swap anything. Call it the Google method.
Sun could have had a compelling desktop-to-datacentre story, all Solaris, on Intel PCs all the way to SPARC near-mainframes. But they succumbed to short term thinking and killed the goose that laid the golden eggs. The alternative to the Oracle acquisition was for them to go the way of SGI...
I'm trying very hard to be as neutral as possible (as opposed to "Oracle bad. Raaaaar!"), but I've been part of the HPC community for a long time, and Oracle has done their best to tell us that we don't matter at all to them (Grid Engine, Lustre, etc.). I understand that the margins on HPC aren't great, so I don't necessarily fault them, but it leaves a very bad taste. That's just one community of many who feel like they've been wronged by Oracle.
You say that they've produced something useful; can you say a little more about what Oracle has specifically produced with respect to Oracle Linux? My strong impression is that they've taken 99.9% Red Hat's hard work, dumped their own 0.1% on top (OCFS2, some IB enhancements, etc.), slapped a big price tag on it and then exclaim, "look what we made!"
Removing that price tag doesn't change much. I'm open to being reasoned with, though, so please tell me why I'm wrong.
There's a ton to talk about in the "Why Oracle Linux and not RHEL" department, but I'm not trying to sell you anything, so I want to set that aside for now. (I have to say, though, even the "big price tag" is significantly smaller than RHEL's).
Instead, I want to look at the following case: you're a sysadmin running CentOS. You don't pay anyone for support, and you have no intention of ever doing so.
There, I think the benefits are, in brief, the things you're able to achieve when you have large-scale resources: timely releases and better QA, while maintaining 100% RHEL compatibility. Basically "what you love about CentOS", minus "what you hate about CentOS".
Useful it may be, but after more than 10 years of having to work with Oracle products and consultants, the first thing that jumped at me was
...doesn't Oracle Linux cost money? A: Oracle Linux support costs money
That "support" word, right there, is the thing that makes me stay as far from Oracle as I can. It's like "Dude, here's the software. Have it, it's cheap/free." When things go wrong you get stung for exorbitant support/consulting fees, because, hey, you're tied in. With nowhere to go.
Sadly, too many organisations still go by the mantra of "The answer is Oracle. Now, what's the question?". That's no basis for a business case.
And yes, this is a rant, and I do have an axe to grind. I'm sorry if that offends (not my intent).
I'm totally with you on fear of lockin, but for what it's worth, this particular case is basically the opposite of lock-in.
You can, at any time, switch away to CentOS, Scientific Linux, or Red Hat (if you're willing to write the check) and not have to totally reinvent your stack from scratch, since they're all binary-compatible.
Your plug for faster updates, for example: Security updates for Solaris used to be free.
So while I know that Oracle is large enough to host many different philosophies in many different departments, what Oracle has done to the former Sun departments really makes me wary of trusting any free offers by Oracle.
Exactly right. I know of several mid- and large-scale companies that replaced (or are replacing) all of their Sun Solaris infrastructure as soon as they heard Oracle acquired them, because of their bad experiences with Oracle.
And from a technical perspective, if the OS is managed anything like the database or client tools, I wouldn't touch it with a 1000ft pole.
I think you did a good job with the vibe, although it was somewhat odd-feeling coming from an oracle.com site.
It did make me double-check the URL bar with the "If you find yourself needing to buy support, have fun reinstalling your system with RHEL before anyone will talk to you". But is perfectly valid.
The problem with Oracle Linux is that it's from Oracle. And I just don't trust them at all. I know I can revert back to CentOS or SL or whatever, but that's a pain. Trust is the problem.
You have a steep hill to hike, but it sounds like you may be cut out to take that on.
I agree, I didn't see anything wrong with the press release. The negative comments here are based on Oracle's long-term reputation. In the wake of the Sun acquisition, they've publicly mismanaged (Hudson) or outright destroyed (OpenSolaris) open-source communities.
I think it's going to take a lot of work for Oracle brand to gain any trust in the eyes of Linux nerds.
You did a good job of the "for linux nerds, by linux nerds" thing. However, when you work for Oracle writing as a linux nerd just comes off as insincere. Your employer is actively working to hurt the linux community, and your continued employment implies your consent of their actions. Your target market needs to be other evil companies that don't care what Oracle does in the courtroom, not linux nerds. "Our product is good" is not a sales pitch that matters right now, because we don't care about your product. It may technologically be the best product for my use case, but that is irrelevant if i don't trust the company behind it. You can call it zealotry if you want, i call it trust and pragmatism. I'm not going to build a business on top of a product that might be used next week as a tool to sue me or my friends. This was the only deciding factor when I chose Postgres over MySqL earlier this year for my company.
If you want to market to linux nerds, you need to change the perception of the company, because that's what matters most right now. Stop being evil, and then we'll give your products a chance.
In all honesty I can understand where you're coming from. However, there is a discrepancy as far as I'm concerned between what you're selling and how you're selling it. In this case you are trying to sell quicker updates, better stability and performance to mostly enterprise customers, with a "we all like to geek around on linux" attitude. If you were trying to sell me on using Oracle Linux for my own development PC the tone would have been quite fine, but for your target audience it seems slightly out of place.
Furthermore, if you are indeed correct and Oracle Linux is something useful that a lot of people will benefit from I seriously doubt that zealotry will stand in the way of its adoption. I guess we will have to wait and see.
Honestly, my target audience with this page is precisely Linux enthusiasts.
Presumably the enterprise customers are all either running RHEL or are already buying Oracle Linux support contracts, and wouldn't be interested in "Oh, I could just run this CentOS-like thing for free".
So, my question is: What's the story on ZFS and DTrace?
I realize I could answer my own question with a little searching, but the main reason I would consider an Oracle branded Linux distribution would be for a reasonably 'official' way of getting some of what made Open Solaris compelling. Whatever the story is, I think my curiosity is predictable and common enough to warrant a mention in your article.
I should note that I am completely uninformed, and only have a passing interest in these things. What little I know is second-hand and years out of date. Although I just did just do a 'sudo yum search zfs' on RHEL 6, which is the only search I should need.
My understanding is:
There's a DTrace beta for Linux. There are no plans to port ZFS to Linux (but, Oracle is investing a lot in making btrfs kick ass).
I'm not looking for a 'kick ass' file system... ZFS does have some features I am interested in though, and I would appreciate having it made available to me somehow.
I was going to leave you alone about the 'messaging' aspects of your post, but now I'll point out that your reply exhibits a similar tone.
Or pragmatism borne out of watching Oracle be a slimy, evil steaming pile of douchebaggery that has only been recently surpassed by Apple in the corporate warmongering department. It's like CentOS, with a heaping helping of evil in every bite.
Unfortunately, I don't think SL holds up in a in all cases. CentOS (and Oracle) strive for binary compatibility. I've had cases where I needed to run vendor supplied binaries which were sometimes buggy on a non-centos EL system. Yeah, it sucks that the vendor only supplies binaries, and only tests against specific RHEL release points, but that's life sometimes.
Don't get me wrong, Scientific Linux is great, but it's not for everyone.
Yes, the priorities of SL are different from CentOS, and binary compatibility isn't quite as high on the list, from what I can tell.
But, so many people are only using Open Source software, and just want a really popular, really stable, really long lifecycle Linux distro for their servers. That's the boat I'm in. I don't currently run anything that I can't get in source form on our servers, so I don't have any concerns about binary compatibility.
That said, both of our products that specifically support SL and have binary packages in their repos (Virtualmin and Cloudmin) use the same yum repository of binaries across RHEL, CentOS, and Scientific Linux. We've never had any bug reports caused by that sharing of binaries. So, binary compatibility is pretty good in our admittedly limited experience (we only provide a handful of binaries).
Yeah, Oracle's brand/goodwill is completely dead if we're talking about consumers. However, does this effect business decisions at all? Any paid thing is aimed at businesses, and the business doesn't care about open source/ethics issues in the same way.
Which business doesn't care? My business does. I know a lot of other businesses that do. It's time to alter the idea that corporations should be free from ethical concerns. Any Open Source supporter working within a company should be helping that company make ethical decisions about software. I'm not suggesting "never buy commercial software" here, I'm just saying that discouraging dealings with demonstrably unethical companies like Oracle is worth doing and it makes business sense.
That's the thing that always kills me about the mindset that ethics don't matter, when you're talking about business decisions. Why would you believe that a company that has repeatedly screwed customers and competitors won't screw you when there's even modest economic benefit to doing so? History indicates you will be screwed, at some point, by Oracle, if you rely upon them. And, by supporting them, you're supporting the ongoing damage they do to our community and the tech industry in general. That's lose-lose.
So, doing the right thing has a pretty good business case.
By the way, Scientific Linux isn't perfect when it comes to these delays either. I wanted to make a similar graph plotting SL, but laziness prevailed. (I crawled the RHEL, CentOS, and OL mailman archives to get the data, but SL uses listserv.)
It's just one data point, but:
Red Hat released RHEL 5.6 on January 13, 2011, but Scientific Linux didn't release SL 5.6 until June 21, 2011
If you run CentOS, it's probably because you're running something that wants to run on RHEL and isn't re-compilable for another platform because it isn't free software. Such as, say, Oracle.
CentOS has for years been the go-to for running Oracle without paying anyone for the OS (RedHat EL or Windows). Now that Oracle has it's own Linux, it makes perfect sense for them to provide a free path into that offering.
I just want a server OS that is stable as hell, has a long lifecycle, and won't break when updating. None of those are quite as true of Debian and Ubuntu (though Debian and Ubuntu have their own benefits, they just don't match the priorities I have for a server OS), and I prefer yum to apt-get, anyway.
And, most of our customers think the same way. CentOS is, by far, the most popular OS for running Virtualmin/Webmin/Cloudmin. Ubuntu has been gaining pretty fast, particularly now that LTS releases have become relatively stable (6.06 LTS had some awful breakages, 8.04 has some minor breakages, 10.04 and 12.04 seem to be pretty stable so far), but RHEL-based systems still have a strong lead in web hosting.
For people already having to sign a deal with the devil, using Oracle may make sense. But, I don't have that requirement. Everything on all five of our servers works fine on CentOS, RHEL, and SL (and Ubuntu, Debian, etc.).
There's a lot of reasons not to use Oracle Linux, most of them non-technical. Oracle is simply not an ethical member of the Open Source community, and if you trust them, they will screw you, some day, some way. Red Hat may have their flaws, but they've never sued over patents and they've never attempted to destroy competing projects or companies through legal threats and bullying. CentOS may be slow to jump on updates sometimes and to get out new releases, but at least it's a good Open Source citizen.
As an aside, if you want a faster moving RHEL rebuild that has paid developers working on it, you might try Scientific Linux. It is built by CERN and Fermilab, and tends to be very solid and fast to update and invisible (i.e., I don't think about it, at all, and it Just Works). We switched from CentOS to SL back when CentOS 6 was so late being released; couldn't be happier with it. We added support in Virtualmin for SL for just that reason...so many of our customers wanted 6, it was worth the effort to add a new OS.
So, yeah, Oracle is gonna have to have a "come to Jesus" moment if they want to participate in the Open Source community. They've got a lot of repenting to do.