Linux Kernel Gets Hot-Data Tracking

[phoronix]

Phoronix: Linux Kernel Gets Hot-Data Tracking

Patches continue to be developed for the Linux kernel that provide hot-data tracking support for file-systems. This hot-data tracking feature may lead to performance improvements when dealing with commonly used data on the disk. Currently this feature has been developed with Btrfs in mind, but could be adapted to other Linux file-systems...

http://www.phoronix.com/vr.php?view=MTIwMzY

[Markore]

Finally something ZFS is having for 6+years in production on Solaris (Illumos and distributions, like Smartos, Nexenta, Openindiana)
You add SSD's to speed up all disk operations for all disks in a pool.

Oh yes, right,
there is now FreeBSD and ZFSonLinux that works fine and is in-kernel ZFS you can use in Linux and it works in production.
And many other things nobody heard before peeking to ZFS specifications that is still not cached up to.

[phred14]

Time for another benchmark.

Every so often Phoronix has benchmarked the various filesystems. In those comparisons, btrfs has always looked interesting, but it has also always preformed quite bad for one case or another. I'm still running ext4 for most of my usage, because while it may not be the best performer, it's always up there and hasn't had bad corners for some time now. (I also use xfs for my MythTV video.)

Hot data tracking looks like another reason to be interested in btrfs.

[fritzls]

today in linux. it's possible to do this kind of data traking?
I'm looking for this for a time.

[ssam]

bcache already exists (though is not in mainline). It lets you assign an SSD as a cache device for an HDD (it actually works with any block device).

http://bcache.evilpiepirate.org/

[LasseKongo]

Finally something ZFS is having for 6+years in production on Solaris (Illumos and distributions, like Smartos, Nexenta, Openindiana)
You add SSD's to speed up all disk operations for all disks in a pool.

Oh yes, right,
there is now FreeBSD and ZFSonLinux that works fine and is in-kernel ZFS you can use in Linux and it works in production.
And many other things nobody heard before peeking to ZFS specifications that is still not cached up to.

I don´t think this is the same, ZFS has it´s ARC disks which keeps a copy of frequently accessed blocks, but the data also remains on your
slow spinning disks at the sae time. My understanding of the hot tracking feature in Linux is that it can also be used to relocate hot blocks, like a true
tiered storage solution. Thar would be really cool, especially if it is filesystem independent.
I also use ZFS for a number of things, but it´s inability to modify existing Vdevs and inability to remove Vdevs from a pool is not very impressive, I
think BRTFS will be a better choice in a year or two when it is feature complete.