Benchmarks Of The Btrfs Space Cache Option

[phoronix]

Phoronix: Benchmarks Of The Btrfs Space Cache Option

In early November we delivered benchmarks of EXT4 vs. Btrfs on an early Linux 2.6.37 kernel as our latest round of tests comparing these two leading Linux file-systems. There were some changes in the Linux disk performance with these file-systems using the latest Linux kernel code, but overall it was not too interesting. However, as the Linux 2.6.37 kernel does introduce a new mount option for Btrfs, the space_cache option, we decided to explore its performance in today's article.

http://www.phoronix.com/vr.php?view=15570

[FireBurn]

Have you done any long term analysis of the compress option?

I used it on both my root and home partitions for a while, at the start the speed was amazing, but slowly over the course of a few months it became slower and slower

It even started to effect the way Chrome and FireFox worked as it was taking so long to open their cache / database

Has anyone else noticed this?

I'm just worried that your praising the compress option because it's only been tested on clean installs and it's not being used for prolonged periods of time

[Ragas]

Have you done any long term analysis of the compress option?

I used it on both my root and home partitions for a while, at the start the speed was amazing, but slowly over the course of a few months it became slower and slower

It even started to effect the way Chrome and FireFox worked as it was taking so long to open their cache / database

Has anyone else noticed this?

I'm just worried that your praising the compress option because it's only been tested on clean installs and it's not being used for prolonged periods of time

Nope I also have the compress option enabled and everything is very speedy.

[FireBurn]

Nope I also have the compress option enabled and everything is very speedy.

How long for?

I remember using reiser4 with compression and it used heuristics to determine what should and shouldn't be compressed

[ChrisXY]

Could you compare the results with same configurations but a magnetic hard disk instead of ssd?

[Ragas]

How long for?

I remember using reiser4 with compression and it used heuristics to determine what should and shouldn't be compressed

Since about 2-3 Month

[R00KIE]

I don't know what kind of data the benchmarks use, but I'd like to see how btrfs performs when dealing with movies, audio or other already compressed data or mixed sets of data.

Benchmarks are fine but we are not all using just databases and as FireBurn suggests, it would be nice to see how a filesystem performs after having data written it enough to fill the partition 2x or 3x.

[sal-e]

Could you compare the results with same configurations but a magnetic hard disk instead of ssd?

In this test is used OCZ 64GB Agility EX SSD. I believe it uses older Indilinx controller. The new Agility II uses SandForce SF-1200 controller. By default SF-1200 compresses the data before its written to the flash memory. So during the test suggested by ChrisXY I would love to see inclusion of SandForce based SSD. It will demonstrate how effective the compression option is when compression is done by SSD's controller.

[LavosPhoenix]

My only problem with btrfs is the lack of a fsck tool that actually corrects errors, otherwise btrfs was working great until something went wrong and I couldn't correct errors in it. However, does space_cache deal with the huge about of "wasted space" that a btrfs system has?

[Flyser]

Testing compression with these benchmarks is absolutely pointless, as they usually only write zeros, which can be compressed very efficently. I think you should mention that in the summary!