BFQ I/O Scheduler For Linux Sees Big Scalability Improvement

[phoronix]

Phoronix: BFQ I/O Scheduler For Linux Sees Big Scalability Improvement

Following the work on enhancing the scalability of the MQ-Deadline I/O scheduler, Linux storage expert Jens Axboe has applied similar scalability improvements to the BFQ I/O scheduler...

https://www.phoronix.com/news/BFQ-IO-Better-Scalability

[Danny3]

This guy is amazing!
Congratulations and many thanks to him!

[cl333r]

It's one of those cases that seem too good to be true ... although avg Linux user wouldn't notice prolly.

[MorrisS.]

In which Kernel are these improvements (on both MQ-Deadline and BFQ I/O) scheduled?

[evert_mouw]

In which Kernel are these improvements (on both MQ-Deadline and BFQ I/O) scheduled?

"scheduled"

[browseria]

In which Kernel are these improvements (on both MQ-Deadline and BFQ I/O) scheduled?

Keep your eye on https://git.kernel.org/pub/scm/linux...alds/linux.git, if it shows up there before rc1 is cut today it will be in 6.8. But probably, I wouldn't expect it before 6.9.

[ffs_]

Nice! Now improve Kyber please.

[bug77]

It's one of those cases that seem too good to be true ... although avg Linux user wouldn't notice prolly.

I was going to say the same. A home user won't see much of a change, but this is going to work wonders for servers, which, like it or not, is still where Linux is most widespread. All of a sudden, cloud providers are able to serve the same level of IOPS using half the storage for spreading reads/writes around.
A welcome improvement nonetheless.

[MorrisS.]

"scheduled"

What do you read about "scheduled" word?

[Danny3]

I was going to say the same. A home user won't see much of a change, but this is going to work wonders for servers, which, like it or not, is still where Linux is most widespread. All of a sudden, cloud providers are able to serve the same level of IOPS using half the storage for spreading reads/writes around.
A welcome improvement nonetheless.

Tell that to KDE Plasma users where the UI stutters or even freezes for a few seconds when there's heavy storage IO!
Maybe these kind of changes help a bit in those cases too.