Announcing: ZFS-fuse 0.7.0
#8 — Benchmarking tool
| State | Resolved |
|---|---|
| Version: |
—
|
| Area | Process |
| Issue type | Feature |
| Severity | Medium |
| Submitted by | (anonymous) |
| Submitted on | Dec 10, 2009 |
| Responsible |
—
|
| Target release: |
—
|
Last modified on
Sep 19, 2010
by
Seth Heeren
In order to share results zfs-fuse could do with a defined benchmark tool/process and a defined way of representing it.
Currently, the benchmarks posted on the mailing list are pretty useless as there's no real way of comparing them and are too open to speculation.
Currently, the benchmarks posted on the mailing list are pretty useless as there's no real way of comparing them and are too open to speculation.
Added by
Christian Kujau
on
Jan 30, 2010 12:00 AM
> In order to share results zfs-fuse could do with a defined
> benchmark tool/process and a defined way of representing it.
Benchmarks are just that: running arbitrary programs executing very special operations and the print the results. There are lots of benchmarking programs out there, I'm not sure what value a "zfs-fuse specific" benchmark program would have - as it then is clearly tailored to zfs-fuse and I'd assume that its results would *not* be comparable to other filesystems, hence defeating the whole point in running a benchmark.
> Currently, the benchmarks posted on the mailing list are pretty
> useless as there's no real way of comparing them and are too open to speculation.
<shameless plug>
I've run a benchmark across filesystmes once, including zfs-fuse (listed as "zfs"):
http://nerdbynature.de/benchmarks/v40z/2009-12-22/
However, as with so many benchmark attempts, please be aware of the (points made in the) discussion
started later on: http://bit.ly/8Z66md
</shameless plug>
> benchmark tool/process and a defined way of representing it.
Benchmarks are just that: running arbitrary programs executing very special operations and the print the results. There are lots of benchmarking programs out there, I'm not sure what value a "zfs-fuse specific" benchmark program would have - as it then is clearly tailored to zfs-fuse and I'd assume that its results would *not* be comparable to other filesystems, hence defeating the whole point in running a benchmark.
> Currently, the benchmarks posted on the mailing list are pretty
> useless as there's no real way of comparing them and are too open to speculation.
<shameless plug>
I've run a benchmark across filesystmes once, including zfs-fuse (listed as "zfs"):
http://nerdbynature.de/benchmarks/v40z/2009-12-22/
However, as with so many benchmark attempts, please be aware of the (points made in the) discussion
started later on: http://bit.ly/8Z66md
</shameless plug>
Added by
Christian Kujau
on
Jan 30, 2010 12:04 AM
Issue state:
unconfirmed → postponed
Added by
Seth Heeren
on
Sep 19, 2010 04:22 PM
Issue state:
postponed → open
Closing due to inactivity (there is no capacity for the devs to do this)
Added by
Seth Heeren
on
Sep 19, 2010 04:36 PM
Issue state:
open → resolved
.
