[Novalug] [Off-list] Re: btrfs vs the world ...

[Varol Okan]

understood. you make a couple of good points ( sharp teethed points if I
may add :)

Anyhow I am aware of the development state of btrfs and I may just be a
bit anxious  about having the feature set it  provides. I would love to
see the performance on a local FS come close to ext3/4. I guess I want
all the goodies of ext3/4 + btrfs's. Sorry, I am just a demanding user
:) Oh and it would be waaaaay cool if my data would be 'save' too ;)

I remember way back waiting for write access the ntfs to the point where
it became available but so late it did not matter to me anymore. I just
wish ( as a Christmas present ) that btrfs will be ready for prime time
before it reaches the same stale state.

Varol :)

Bryan J. Smith wrote:
> It's an "eye opener" for me, but not in the way you might think.
>
> 1.  Anyone who evaluates filesystems based on performance (let alone
> distributed filesystems) are going after the wrong metric.  It's way, way
> down on my list of considerations.
>
> INSIGHT:  There's not a week that goes by where people don't remotely
> understand the difference between local and distributed filesystems.
> Instead of questioning their own usage (let alone classification/need of
> their data to be distributed), they blame the filesystems.  This is also the
> case with other usage (not just local v. distributed).
>
> 2.  EXT4 development has been funded by Red Hat, with the major
> knowledgebase from the same -- Tweedie (probably the man who
> most made Ext2/Ext3 "not suck") and Sandeen (SGI XFS developer,
> and help architect and re-wrote a lot of the VFS in kernel 2.5+ that
> all filesystems benefit).  So why is Phoronix benchmarking Ubuntu?
>
> INSIGHT:  There's not a week that goes by where the tales of "oh, I
> use Ubuntu at home, so I brought it in for our servers" -- not even Ubuntu
> LTS, but leading-edge Ubuntu -- and then people wonder why they not
> only have issues, but the "knowledge" is severely lacking in not just the
> "free" realm, but the "paid" realm as well.
>
> I.e., when it comes right down to it, for enterprise usage, the companies 
> that actually develop most of the code can do the most for you when you
> run into the implementations and cases _outside_ home usage or simple
> web server usage, and you start hitting that "enterprise infrastructure"
> usages.  I don't say that as a slant towards Canonical/Ubuntu, but just a
> repeat, sad theme that I see at clients/customers way too much.
>
>
>
>
> ----- Original Message ----
> From: Varol Okan <varokan at movingsatellites.com>
>
> This is an eye opener I think.
> http://www.phoronix.com/scan.php?page=article&item=linux_2632_fs&num=1
>
> I was planing in playing with btrfs, to see performance but after the
> Phoronix test results, I guess this would be wasted time ( wrt
> performance anyways ).
>
> I like the feature list of btrfs  a lot and it might make a lot of sense
> for certain applications ( I am especially thinking about the snapshot
> capabilities ). So I hope that the performance in general can be
> improved to come at least close to ext3 / ext4.
>
> Is anyone here using, testing, or playing with btrfs ?
>