Slashdot Mirror
Red Hat Hires CentOS Developers
rjmarvin writes "Karanbir Singh and a handful of other CentOS developers are now full-time Red Hat employees, working in-house on the CentOS distribution with more transparent processes and methods. None of the CentOS developers will be working on Red Hat Enterprise Linux. The CentOS project would become another distribution and community cared for by Red Hat, like Fedora, and Red Hat CTO Brian Stevens says the company is planning its future around OpenStack, not just Linux."
Nah. Oracle would rather kill Solaris than let that happen.
and not embrace and extinguish. Kudos to Redhat and CentOS.
Buy them! Or hire them like in this case..
I've always wondered what RH could do about CentOS. It was obvious that RH wasn't all that happy with CentOS, at least at first. With CentOS having to refer to "the up-line vender" and removing all the RH references and graphics it has always seemed to be the Red Headed step child.
So, does this mean RH has embraced the concept of CentOS, where "free is free" to download?
"File to fit, pound to insert, paint to match" - Aircraft Maintenance 101
I've wondered why Google hasn't put out their own Linux distro for the mainstream audience. They could have eaten Microsoft's lunch if they put that out when they went to Vista. It would have cost Google next to nothing to fork a common distro like Ubuntu and slap their name on it, and by incorporating Google search and Chrome (or Chromium) into it, they could have increased their ad revenue. I guess ChromeOS is sort of like that, but I was imagining a full desktop OS.
Tic-Tac-Toe, Global Thermonuclear War, and relationships all have the same winning move.
Sadly, I believe you're correct.
Solving Unix problems since 1989...
Given that Redhat is now officially cooperating, I'm not entirely sure why CentOS is still relevant(rather than 'Redhat, RTFM Edition, upgradeable at any time to Redhat, Comes with Support Edition if you buy support'); but I assume that Redhat is focused on the enterprise for a few reasons:
1. Enterprise is where MS, and any other remaining competitors, really turn the screws on pricing. MS doesn't give away Windows Home editions; but only the OEMs know how much those costs, and most buyers aren't considering DIY or buying a 'bare' PC, so the effective cost (among the options they have) is zero. Enterprises, Not. So. Much. MS charges considerably more for 'Pro', and more again for anything server.
2. Enterprises have volume and techies. A home user has, maybe, the nerd kid down the street or something for tech support. They also have a small number of computers. Even a relatively high price, per computer, makes total sense if it avoids any support headaches, and allows those that do come up to be handled by the most common tech support people. Enterprises, though, have enough computers that buying techs rather than 'solutions' starts to become cost effective(plus, their requirements tend to be complex enough that 'solutions' still require techs)
3. 'Desktop'(in the sense of 'consumer') is where a lot of the really nasty hardware churn is. 'Enterprise desktop', 'workstation', and 'server' are all areas where (even if running Windows) IT departments Do. Not. Want. lots of driver/hardware churn, don't want to spend lots of time re-validating configurations, don't want shitty beta drivers, and so on. They are also often satisfied with a smaller variety of hardware, and from vendors who are more likely to build drivers with server and workstation customers in mind. Consumer OS that doesn't support a shitty inkjet released two years after the OS was? Pissed off consumer. Enterprise? Well, we've got some printers that all support Postscript or PCL, a bunch of servers that need NIC and SAS HBA support, and maybe some workstations with fancy graphics cards.
(As for Google and consumer Linux, it's a matter of taste whether you say that they already have, or that they never will: Android and ChromeOS are both Linux-based, neither have more than the slightest relationship to traditional linux/unix userlands. Is Google throwing its weight behind consumer Linux, or using embedded Linux as a cheap and easy way to boot a Google userland?)
CentOS is the freebie that anyone can use - but that nobody is under any obligation to provide support or patches for. This means that small companies who are waiting to grow before buying proper RHEL, can still use the software, though they can't file bugs or get a support hotline. But it also means that CentOS can be used for anyone training in skills for "Enterprise grade" Linux can get their feet wet on a system that is already in use in industry. When the time comes to work with Linux in a real business environment, they've a head start on those who chose systems closed to non-customers.
Why does this matter to RedHat? The more people whose yardstick and gold standard is RedHat-related technology, the better; and ensuring all you can do on the derivative can be done exactly the same way on the commercial (down to the version of a command, the dot in a package name and the quirks of the brand) goes a long way to provide this promise.
How does this benefit RedHat if CentOS is given away for free? CentOS is RedHat's technology already in the hands of the client. But having the software is one thing - having access to support, formal enterprise training offerings, consultancy services and a dedicated rapid response for business-critical bugs is vital in business. Once the small company who could not afford RHEL becomes big, suddenly they are aware that they are on systems that RedHat knows perfectly, and migrating from CentOS to RHEL is painless - being systems different only at branding level. Migrating to anything else, even to SUSE Linux for Enterprise or Oracle's Linux (the latter being a part-clone of RHEL), becomes more involved. CentOS really now is RHEL.
Indeed, the good karma from being seen helping the community is peanuts compared to the advantage the offering of an easy transition and self-trained fans and already-committed users brings.
-- "Simplicity is prerequisite for reliability." --Dijkstra
Nah. Oracle would rather kill Solaris than let that happen.
Good riddance... Solaris is nothing special or unique, at least these days. It's best feature was that it ran on Sparc based hardware, which when Sun Micro Systems was in it's heyday said "Rock Solid reliability". So if you wanted something to run for a decade or two, you purchased Sun hardware which locked you into Solaris. Now days, who cares about Solaris? Running Solaris on X86 hardware it is pointless because it buys you nothing in reliability while costing you in obscurity. Just go to a stable Linux distribution.
The ONLY reason you field Solaris now, is if your customer demands it or your legacy application is not easily ported. The one possible exception to this might be if you are putting up an Oracle cluster.
Solaris is going to die... It probably should too.
"File to fit, pound to insert, paint to match" - Aircraft Maintenance 101
Really, you think Google could have just slapped their name on it and all the reason Linux hasn't taken off on the desktop would go away? The other part is that Google isn't very interested in giving you a local solution, that doesn't give them any data nor a hook to google services. And the third reason is that it would be too easy for third parties to strip off the Google bits, despite Android being open source they have very strong incentives to keep OEMs from shipping "bare" phones without all the Google services, the Play store and so on which they wouldn't have for a Linux distro.
Live today, because you never know what tomorrow brings
Really, you think Google could have just slapped their name on it and all the reason Linux hasn't taken off on the desktop would go away?
Not quite what I said. But if Google put their weight behind marketing a Linux OS at a time when people were unhappy with Microsoft's Vista and/or recently with Windows 8, yes I think Linux on the desktop could have taken off more so than it is now.
The other part is that Google isn't very interested in giving you a local solution, that doesn't give them any data nor a hook to google services.
You mean like ChromeOS, with Google search built in with Google apps throughout? Why not do the same with a full desktop OS?
And the third reason is that it would be too easy for third parties to strip off the Google bits
Android is open source, but not all of it. Google could easily market a Linux distro with some proprietary software integrated in, that most users would be able to/want to remove. You can get a Google-free version of Android, but you have to unlock the bootloader, load recovery, root, and install a new ROM; not something most users are going to do. Why not do something similiar with a full desktop Linux-based OS?
Tic-Tac-Toe, Global Thermonuclear War, and relationships all have the same winning move.
Let me take a second to applaud Red Hat for doing this.
This is why they own the Enterprise Linux market.
Their thinking, in a nutshell, is this:
Give the software (CentOS) to small companies.
Get kids right out of college using it to build their home servers.
Get everyone comfortable with CentOS/RHEL.
When it is time to buy, they will buy RH. Simple.
Here in NYC, Linux jobs are 99.99% RH/CentOS.
Because CentOS is free, anyone can download it and test
it. No disabled features, nothing. You want a job in Wall St?
Download CentOS, sit down and learn the thing and then
you WILL get a job! I guarantee it!
Microsoft, Oracle, Apple take note: This is how you own
a market. Not by squeezing every penny out of your
customers.
That's why Apple will never break into the Enterprise
market. This is why Microsoft has lost the Enterprise
market and this is how Oracle will fuck off and die soon
(hopefully).
Personally, I was a Slackware guy, for my home machines,
but CentOS has won me over. Now, it is the only thing I use.
One more thing: I work in Wall St. and I use RH/CentOS
every single day.
Red Hat, you guys rule. I salute you! Rock on!
It's called Windows XP.
Support ends in 2 months, so good luck using after that. And it doesn't and hasn't helped when people are shopping for a new machine.
Tic-Tac-Toe, Global Thermonuclear War, and relationships all have the same winning move.
Nah. Oracle would rather kill Solaris than let that happen.
Good riddance... Solaris is nothing special or unique, at least these days. It's best feature was that it ran on Sparc based hardware, which when Sun Micro Systems was in it's heyday said "Rock Solid reliability". So if you wanted something to run for a decade or two, you purchased Sun hardware which locked you into Solaris. Now days, who cares about Solaris? Running Solaris on X86 hardware it is pointless because it buys you nothing in reliability while costing you in obscurity. Just go to a stable Linux distribution.
The ONLY reason you field Solaris now, is if your customer demands it or your legacy application is not easily ported. The one possible exception to this might be if you are putting up an Oracle cluster.
Solaris is going to die... It probably should too.
Actually, Solaris had several good features. It had much better resource management than Linux does, and the introduction of Solaris Zones allowed a high-performance VM environment while minimizing the amount of gratuitous replication of resources. Basically, sort of a chroot jail + COW filesystem with service level controls.
Sadly, however, the Solaris admins where I worked were never trained to take advantage of those features, so eventially Solaris got booted in favor of lots of Linux and Windows boxes.
Full disclosure... I've been a Solaris Admin, off and on for years who has only been briefly involved with using zones
Your comments show this.
The zones idea is roughly equivalent to chroot (or schroot in some use cases) on Linux. So if you like zones, you can do almost the same thing on a Linux box.
No, it's not. Chroot is a (mostly) completely useless mechanism for security and isolation.
With zones, you can set up the zone with a completely different IP, with different firewall rules and even routing tables, and give some access to the root account on that zone, and not have to worry about them breaking out of the zone or affecting the hosting system (because you can put memory and CPU restrictions on the zone so it doesn't eat up system-wide resources). You can have dozens of zones on one hosting machine and the overhead of this "virtualization" is (IIRC) less than 3%.
FreeBSD's jails is the closest equivalant (having inspired zones). With jails you can also give out the root account and not worry, and even have set-UID binaries.
Put a set-UID binary in a chroot space (or even a Linux container) and all security is gone.
Zones/jails are nothing like chroot: the former can actually be used securely (where do you think the first VPS system came from? FreeBSD jails), while the latter is a nice speed bump before being broken out of.
Replacing systemd with smf and btrfs with zfs would be fantastic. Solaris 10 and OpenSolaris did so many things just right and Linux so wrong.
"I use a Mac because I'm just better than you are."
Nothing except ZFS, DTrace, crossbow, a fully featured SCSI target stack, and on solaris an extremely robust partitioning infrastructure. Here we are 9 years later and Linux still doesn't have a good answer.
Well, the example I gave was the network. Yes, chroot does help seperate groups processes (although it's important to understand the seperation is insecure by design, that is any root process can easily escape a chroot environment and have access to the main file system, which means you're limited in what types of process you can run if you want a chroot'd environment to be "secure"), but there do exist operating system functions that are not and never have been accessable via the filesystem.
The example I gave, networking, is part of the reason for LXC. LXC in turn is inspired by (and developed by the same developers as) an independent project called OpenVZ, which is used by many hosting providers to provide a kinda poor-man's VPS service. Each LXC domain has its own virtual network device, with its own IP address. Each LXC domain has a secured filesystem, with no risk of a "break out".
As I said, my understanding is that zones is similar in concept to LXC, not chroot. chroot only covers a small part of what zones does. As an aside, chroot is actually a feature of every version of Unix since version 7 (late seventies/early eighties), if zones and chroot were equivalent, it'd be a redundant feature.
You are not alone. This is not normal. None of this is normal.