Oracle's Unbreakable Enterprise Kernel R5 Running On Linux 4.14

[phoronix]

Phoronix: Oracle's Unbreakable Enterprise Kernel R5 Running On Linux 4.14

Oracle's current Unbreakable Enterprise Kernel (UEK) release is making use of Linux 4.1 that is definitely showing its age... Fortunately, for those using this Oracle spin of the Linux kernel on their RHEL-clone Oracle Linux have UEK Release 5 in the pipe that is transitioning to Linux 4.14 LTS...

http://www.phoronix.com/scan.php?page=news_item&px=Oracle-UEK-R5-Linux-4.14

[RealNC]

The name annoys me to no end.

Someone should offer a kernel and name it "unbeatable kernel that has shown to be superior to everything else on the market" in order to abuse media coverage. Like Oracle does with this stupid "unbeatable" nonsense.

F off.

[grok]

"Unbreakable" they say. Like a Soviet fridge that has no feature, weighs a ton and gobbles electricity but dates from Brezhnev and still works after the USSR has ended.
They do have a competitor called "NonStop", isn't that pretentious? (I don't know how marginal that is. Maybe five or six banks or airlines paying $10 million a year to run their 1980s stuff?)


Of course it says "Oracle" which is a much bigger problem that any epiteth.

[Spooktra]

OK, I'll bite, what exactly is an "Unbreakable Kernel"? I guess the answer is closely related to "How do you brake a kernel"? And if the kernel is "unbreakable" what's the point of upgrading it?

[uid313]

It's basically RHEL, much like CentOS.
Better go with Red Hat than Oracle.

[starshipeleven]

OK, I'll bite, what exactly is an "Unbreakable Kernel"? I guess the answer is closely related to "How do you brake a kernel"? And if the kernel is "unbreakable" what's the point of upgrading it?

It's marketing-speech, it probably relates to how you can't break the linux kernel from userspace, so whatever you throw at it will be handled better than on Windows.

Of course this is nonsense if you start adding drivers, as Linux will easily crash and burn if the driver itself is shit, which usually does not happen with open drivers as they are much higher quality than most closed ones.

As a matter of fact, the only kernel that can be called "unbreakable" is a microkernel design, as there the drivers can crash and burn all they want, but the kernel itself is more compartimentalized.

[UpsetingFact]

OK, I'll bite, what exactly is an "Unbreakable Kernel"? I guess the answer is closely related to "How do you brake a kernel"? And if the kernel is "unbreakable" what's the point of upgrading it?

Basically a kernel that ensures there is no possible remaining bugs.

Free softwares are "Realease often, release early"-trigger happy (or rolling releases), and that's utter shit.
So, Oracle is doing true stable release (what every software should follow) for serious uses that won't need to wait for the next version to fix bugs, while bringing possible new bugs.

Granted Oracle's naming is shit, but if people would use "rolling" instead of fake "stable" for their releases, they wouldn't have to make such shitty name.
(Note that I hate Oracle as hell, except for killing Sun Microsystems thus slowling killing Java; and doing real stable release).

[Vistaus]

As a matter of fact, the only kernel that can be called "unbreakable" is a microkernel design

And kernels written in Rust :P

[torsionbar28]

"Unbreakable" they say. Like a Soviet fridge that has no feature, weighs a ton and gobbles electricity but dates from Brezhnev and still works after the USSR has ended.
They do have a competitor called "NonStop", isn't that pretentious? (I don't know how marginal that is. Maybe five or six banks or airlines paying $10 million a year to run their 1980s stuff?)

The Tandem (later Compaq, then HP) "NonStop" has earned its name. I used to work on these years ago. The hardware and software level redundancy is insane, nothing else on the market even comes close. Everything in a NonStop server is "quad redundant". The CPU's, the RAM, the I/O, everything, is mirrored 4x. Even the hard drives were special and unique to the NonStop servers because they were low level formatted at 514 bytes per sector instead of 512 bytes. This is because the NonStop OS and filesystem wrote a 2 byte checksum value to every hard drive sector, thus completely eliminating the possibility of silent corruption. Nobody else was doing this stuff, NonStop was unique. This is not "1980's stuff" as you say, but it's so exotic, I don't blame you for being ignorant of it.

You can literally walk up to the NonStop server and start yanking out chips and circuit boards and it just keeps running, no interruption. When I worked at Compaq, they did a demo video of a NonStop server running a variety of workloads, and they had someone fire a 50 caliper bullet through the side of the chassis. They showed in slow motion, the bullet piercing all kinds of CPU and memory boards inside, and the bullet went completely through, exiting out the other side of the chassis. There was no interruption or delay or pause of any kind in the workload processing. Hence the name, "NonStop".

The nice side effect of such extreme redundancy and resilience is that you can upgrade literally every part in the box with zero downtime. When newer faster CPU's or more dense RAM or faster I/O backplanes became available, the field technician could replace all of it with no outage.

Fun fact: 14 of the 15 stock exchanges in the world run on NonStop, and 100% of nuclear power generation in the US runs on NonStop, and 70% of all non nuclear power generation in the US does too.

Of course it says "Oracle" which is a much bigger problem that any epiteth.

Agreed.

[jacob]

As a matter of fact, the only kernel that can be called "unbreakable" is a microkernel design

So you move all the breakable parts into userspace. You end up with an OS that is every bit as breakable as before, except that you technically don't call it the kernel.