Block Updates Land In The Linux 4.16 Kernel

[phoronix]

Phoronix: Block Updates Land In The Linux 4.16 Kernel

The block subsystem updates have now landed in Linus Torvalds' Git tree during the first full day of the Linux 4.16 kernel merge window...

http://www.phoronix.com/scan.php?page=news_item&px=Linux-4.16-Block-Updates

[Adarion]

Does anybody really use SMR HDDs for anything but archiving? I have yet to hear fond words (that aren't marketing) about this technique.

[caligula]

Does anybody really use SMR HDDs for anything but archiving? I have yet to hear fond words (that aren't marketing) about this technique.

They're not that much slower, at least for home users. Lots of random IOPS will kill the perf, but they can handle stuff like copying masses of files and extracting archives just fine. Assuming the OS buffers the I/O first.

[jpg44]

Good to see performance improvements as Linux is an industrial strength operating system used in the most demanding environments.

[coder]

They're not that much slower, at least for home users.

Right, home users. Just don't dare try running something like a decent sized database, with lots of updates, or even a rather small number of VMs.

SMR is fine to replace tape, optical discs, and basically anything that's highly read-biased. But it doesn't take many random writes before you're swamped.

[Guest]

There isn't any "breakthrough" new block features for Linux 4.16 but the block optimizations are notable and I look forward to benchmarking that shortly.

There are others as well:

https://lkml.org/lkml/2018/1/29/230

In my own testing this series significantly helps performance with small I/O sizes. I also got this email for Christmas this year from the kernel test robot (a 244% r/w bandwidth improvement with XFS over DAX, with 4k writes)

[starshipeleven]

Does anybody really use SMR HDDs for anything but archiving? I have yet to hear fond words (that aren't marketing) about this technique.

are SMR hard drives supposed to be used for anything but archiving?