[ale] BTRFS - used it?

[Ed Cashin]

Just a note that I had to use a 2.6.30.4 kernel yesterday and
so tried out nilfs2, which promises to be a great filesystem, as
this article argues:

  http://www.linux-mag.com/cache/7345/1.html

It has the features that the new filesystems tout (but not quotas
or fs growth yet), performs better than or close to the levels of
its peers, and has a very interesting automatic checkpointing
feature that is more or less free.

You can make any checkpoint into a snapshot so that it doesn't
get garbage collected later, and then you can mount it read only
while the "live" version of the filesystem is still mounted read/write.

It is a log structured filesystem, so it *might* handle writes better
than other filesystems.

To me the most promising thing is that even though it warned
me that it's in the development stage, and it just appeared in the
mainline kernel, it seemed rock solid when I tried it.  Not even a
head scratcher to overcome.  (As you'd expect, I needed to get
the userland tools to mkfs and mount, etc.)

It seems like an interesting alternative to btrfs and ext4.

They plan to support extents.  We'll see how that goes.  If it
continues to deliver on its promises it could replace XFS as the
filesystem I recommend to customers at work.  XFS is pretty
fast and grows to huge capcacities, but the development churn
seems to keep its stability from being a sure thing when you try
a new kernel.

-- 
  Ed Cashin <ecashin at noserose.net>