Distributed Linux OS wizards CoreOS release first commercial product

A company with a mad plan to right the wrongs perpetrated on the world by traditional operating systems has released its first commercial product after taking in $8m in venture capital funding. CoreOS made the commercial version of its CoreOS open source operating system available for sale on Monday, before relaxing on a …

[John Sanders]

Please...

Use our services and disconnect your brain sir.

[Dr Who]

Having skimmed their website, at first glance this looks to be very nice stuff. In many ways it helps address the main criticism of running your stuff on cloudy infrastructure, namely the dependence on a single third party whose operations a opaque. You can run CoreOS instances all over the place, say Amazon, Rackspace, Google, and on your own hardware in your own DC and use the whole lot as a single cluster. With Docker in the mix, I'd say this is well worth keeping an eye on.

[Anonymous Coward]

> The tech uses three key technologies: etcd...; Docker...; and systemd...

systemd? I thought they were trying to cut down on the bloat!

[chrismeggs]

Wow!

Certainly ticks my boxes, if not for this initial release, but for the implied direction. One to watch.

[Anonymous Coward]

updating it automatically

>>hopes to keep the OS up to date: by updating it automatically for its users

Not sure that's a good idea. Like giving out your root password to a chimp?

[Anonymous Coward]

Re: updating it automatically

"Like giving out your root password to a chimp?"

No biggie - the Microsofties go all cross-eyed when they see a Bash prompt anywho...

[Paul Crawford]

Update all at once?

That makes sense for laptops, etc, which get powered up/down irregularly so having two root partitions that toggle once safely updated makes sense.

But for a main OS then every update means a reboot. OK, that might be part of the overall system design and fail-over strategy for cloud use, but you can say goodbye to updating just a broken library, etc, with the rest of the machine (or at least that kernel) and running processes keeping going.

[htq]

Re: Update all at once?

Reboot after every update? We're not talking about Windows server here ;-)

Patching the running kernel might mean a reboot - but not necessarily if you're using Ksplice.

[cyberelf]

Re: Update all at once?

"The update strategy is defined in cloud-config ..The number of machines allowed to reboot simultaneously is configurable via a command line utility"

[Anonymous Coward]

Re: Update all at once?

Paul, you appear to be thinking in single-machine terms. What part of distributed did you not understand?

[Paul Crawford]

Re: What part of distributed did you not understand?

Maybe the bit when I said: "that might be part of the overall system design and fail-over strategy for cloud use"

Either way, you still have to migrate running jobs off a given machine/kernel in order to have it updated when traditional Linux could often be kept going.

[frank ly]

Updates by outsiders

If you're worried about the 'quality' and suitability of updates, can't you load them into your own repository and then decide which ones you'll allow onto your system after doing test and checks on them?

[Michael Wojcik]

Still thinking too small

OS-as-a-service

Bah. Let me offload all my costs. Someone bring me enterprise-as-a-service, where they host my entire business and I just make grand statements to the investors and press.

Actually, that sounds like too much work. Sign me up for profit-as-a-service.