blu.org  wiki

ext3cow FS?

View: New views
10 Messages — Rating Filter:   Alert me  

ext3cow FS?

by Scott R. Ehrlich :: Rate this Message:

Reply to Author | View Threaded | Show Only this Message

I just found an article comparing Ubuntu Linux (7.04) to Vista, and one of the
areas they compare is backups.   In that section they mention ext3cow, which
was the first time I'd heard of it.

What are people's insights of ext3cow?   I haven't had to recompile a kernel in
ages - it seems the default ones, for the most part, for Ubuntu and CentOS,
anyway, are fully robust to handle whatever I throw at them.

So how does something like ext3cow for some kind of backup/restore fair?
What other similar tools do people use?   How about for systems that are
actively in use that I may opt to insert this functionality, ideally without
rebooting, but if needed, I would.

Thanks.

Scott

--
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.

_______________________________________________
Discuss mailing list
Discuss@...
http://lists.blu.org/mailman/listinfo/discuss

Re: ext3cow FS?

by Tom Metro-12 :: Rate this Message:

Reply to Author | View Threaded | Show Only this Message

Scott R. Ehrlich wrote:
> I just found an article comparing Ubuntu Linux (7.04) to Vista...

Link?


> In that section they mention ext3cow...

Link?
Oh, here it is: http://www.ext3cow.com/  (Thanks Bill)
Also : http://en.wikipedia.org/wiki/Ext3cow


> ...which was the first time I'd heard of it.

Me too. Thanks for mentioning it.


> I haven't had to recompile a kernel in ages - it seems the default
> ones, for the most part, for Ubuntu and CentOS, anyway, are fully
> robust to handle whatever I throw at them.

And your point is? Are you implying use of ext3cow requires compiling a
custom kernel? The blog posting Bill linked to:

http://cmynhier.blogspot.com/2007/05/ext3cow-snapshots-for-ext3.html

says:

   One of the design features of ext3cow is that changes to support
   snapshotting were localized to the ext3 code itself... This certainly
   makes life easy, as it can be used with otherwise stock kernels.

It doesn't say explicitly, but I assume the author is implying that it
can be built as a module.


> What are people's insights of ext3cow?

Per this from the blog posting:

   Multiple versions of a file can share data blocks, as long as those
   data blocks haven't changed.

I'd say it fairs better than most other snapshotting technologies
available on Linux. The other options generally operate at the file
level, rather than at the block level. This sounds comparable to what is
done in NetApp's WAFL[1] or Sun's ZFS[2], both file systems designed
with snapshotting in mind. You can do snapshots with LVM, but I'm pretty
sure it isn't as efficient. To quote Wikipedia, "Block-level
snapshotting is almost always less space-efficient than direct file
system support for snapshots."[3]

1. http://en.wikipedia.org/wiki/Write_Anywhere_File_Layout
2. http://en.wikipedia.org/wiki/Zfs
3. http://en.wikipedia.org/wiki/Snapshot_%28computer_storage%29


What I don't see in either the ext3cow site or the blog posting is an
indication of the maturity or stability of the software, aside from the
0.2 version number. I'd have to assume it is still in an experimental state.


> So how does something like ext3cow for some kind of backup/restore fair?

I'd assume it supplements a backup strategy just as any other
snapshotting[3] technology does. It permits you to more quickly access
past versions of a file, and to efficiently store more versions. Of
course that's contingent upon the file system being accessible, and thus
you still need to pair it with traditional backup techniques.


> What other similar tools do people use?

rsnapshot[4] and other similar tools built around rsync have
traditionally been the approach to attaining snapshot functionality on
Linux. I see mention of LVM snapshots some too.

4. http://www.rsnapshot.org/

  -Tom

--
Tom Metro
Venture Logic, Newton, MA, USA
"Enterprise solutions through open source."
Professional Profile: http://tmetro.venturelogic.com/

--
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.

_______________________________________________
Discuss mailing list
Discuss@...
http://lists.blu.org/mailman/listinfo/discuss

Re: ext3cow FS?

by Kristian Erik Hermansen :: Rate this Message:

Reply to Author | View Threaded | Show Only this Message

On Sun, May 4, 2008 at 8:10 AM, Scott R. Ehrlich <scott@...> wrote:

> I just found an article comparing Ubuntu Linux (7.04) to Vista, and one of
> the areas they compare is backups.   In that section they mention ext3cow,
> which was the first time I'd heard of it.
>
>  What are people's insights of ext3cow?   I haven't had to recompile a
> kernel in ages - it seems the default ones, for the most part, for Ubuntu
> and CentOS, anyway, are fully robust to handle whatever I throw at them.
>
>  So how does something like ext3cow for some kind of backup/restore fair?
>  What other similar tools do people use?   How about for systems that are
> actively in use that I may opt to insert this functionality, ideally without
> rebooting, but if needed, I would.

I'll just wait for ZFS on Linux, but it's nice to know ext3cow is an
option too :-)
--
Kristian Erik Hermansen
--
"Clever ones don't want the future told. They make it."

--
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.

_______________________________________________
Discuss mailing list
Discuss@...
http://lists.blu.org/mailman/listinfo/discuss

Re: ext3cow FS?

by Scott R. Ehrlich :: Rate this Message:

Reply to Author | View Threaded | Show Only this Message

On Sun, 4 May 2008, Tom Metro wrote:

> Scott R. Ehrlich wrote:
>> I just found an article comparing Ubuntu Linux (7.04) to Vista...>
> Link?
>

Sorry -
http://www.informationweek.com/news/software/linux/showArticle.jhtml?articleID=199201179&pgno=7&queryText=&isPrev=

or

http://tinyurl.com/5wepbp

Scott

--
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.

_______________________________________________
Discuss mailing list
Discuss@...
http://lists.blu.org/mailman/listinfo/discuss

Re: ext3cow FS?

by Jerry Feldman-2 :: Rate this Message:

Reply to Author | View Threaded | Show Only this Message

On Sun, 4 May 2008 14:35:29 -0400 (EDT)
"Scott R. Ehrlich" <scott@...> wrote:

> On Sun, 4 May 2008, Tom Metro wrote:
>
> > Scott R. Ehrlich wrote:
> >> I just found an article comparing Ubuntu Linux (7.04) to Vista...>
> > Link?
> >
>
> Sorry -
> http://www.informationweek.com/news/software/linux/showArticle.jhtml?articleID=199201179&pgno=7&queryText=&isPrev=
>
> or
>
> http://tinyurl.com/5wepbp
Not a very good article.  He only talks about Konserve which is a KDE
front end to tar.
"One problem is that it doesn't seem possible to filter files to be
backed up; it's everything in the source directory or nothing. Also,
each backup set is complete; the program doesn't have an explicit
option to perform incremental backups. (I also looked at the Keep
Backup system, which had a similar set of options but also many of the
same limitations.)"

He does not mention that tar, rsync, and other command line tools have
advanced filtering capabilities. He also fails to mention other backup
solutions on Linux.


--
--
Jerry Feldman <gaf@...>
Boston Linux and Unix
PGP key id: 537C5846
PGP Key fingerprint: 3D1B 8377 A3C0 A5F2 ECBB  CA3B 4607 4319 537C 5846

_______________________________________________
Discuss mailing list
Discuss@...
http://lists.blu.org/mailman/listinfo/discuss

Re: ext3cow FS?

by Dan Ritter-2 :: Rate this Message:

Reply to Author | View Threaded | Show Only this Message

On Sun, May 04, 2008 at 11:10:14AM -0700, Kristian Erik Hermansen wrote:
> I'll just wait for ZFS on Linux, but it's nice to know ext3cow is an
> option too :-)

It seems likely that you'll be waiting a long time. ZFS is under
the CDDL, which is highly non-GPL compatible, and Sun's patents
suggest that they don't want anyone reverse-engineering it,
either. Nor do I think Sun management has much of a soft spot
for Linux these days...

-dsr-

--
http://tao.merseine.nu/~dsr/eula.html is hereby incorporated by reference.

"Someone is wrong on the Internet." - Randall Munroe

--
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.

_______________________________________________
Discuss mailing list
Discuss@...
http://lists.blu.org/mailman/listinfo/discuss

Re: ext3cow FS?

by Kristian Erik Hermansen :: Rate this Message:

Reply to Author | View Threaded | Show Only this Message

On Sun, May 4, 2008 at 3:58 PM, Dan Ritter <dsr@...> wrote:
>  It seems likely that you'll be waiting a long time. ZFS is under
>  the CDDL, which is highly non-GPL compatible, and Sun's patents
>  suggest that they don't want anyone reverse-engineering it,
>  either. Nor do I think Sun management has much of a soft spot
>  for Linux these days...

Perhaps ;-/  You can already use ZFS on Linux via FUSE, but I would
hope that someone will want it bad enough to reimplement it.  You can
get around the reversing-to-implementation problem using the Chinese
Wall methodology, so I don't see why that would be a problem...
--
Kristian Erik Hermansen
--
"When you share your joys you double them; when you share your sorrows
you halve them."

--
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.

_______________________________________________
Discuss mailing list
Discuss@...
http://lists.blu.org/mailman/listinfo/discuss

Re: ext3cow FS?

by Jarod Wilson :: Rate this Message:

Reply to Author | View Threaded | Show Only this Message

On Sun, 2008-05-04 at 16:08 -0700, Kristian Erik Hermansen wrote:

> On Sun, May 4, 2008 at 3:58 PM, Dan Ritter <dsr@...> wrote:
> >  It seems likely that you'll be waiting a long time. ZFS is under
> >  the CDDL, which is highly non-GPL compatible, and Sun's patents
> >  suggest that they don't want anyone reverse-engineering it,
> >  either. Nor do I think Sun management has much of a soft spot
> >  for Linux these days...
>
> Perhaps ;-/  You can already use ZFS on Linux via FUSE, but I would
> hope that someone will want it bad enough to reimplement it.  You can
> get around the reversing-to-implementation problem using the Chinese
> Wall methodology, so I don't see why that would be a problem...

btrfs 0.14 released just a bit ago...

http://kerneltrap.org/Linux/Btrfs_0.14_Managing_Multiple_Devices


--
Jarod Wilson
jarod@...


--
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.

_______________________________________________
Discuss mailing list
Discuss@...
http://lists.blu.org/mailman/listinfo/discuss

Re: ext3cow FS?

by Kristian Erik Hermansen :: Rate this Message:

Reply to Author | View Threaded | Show Only this Message

On Sun, May 4, 2008 at 5:17 PM, Jarod Wilson <jarod@...> wrote:
>  btrfs 0.14 released just a bit ago...
>
>  http://kerneltrap.org/Linux/Btrfs_0.14_Managing_Multiple_Devices

I forgot about btrfs!  Very nice...
--
Kristian Erik Hermansen
--
"When you share your joys you double them; when you share your sorrows
you halve them."

--
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.

_______________________________________________
Discuss mailing list
Discuss@...
http://lists.blu.org/mailman/listinfo/discuss

Re: Chinese Wall and software patents

by Tom Metro-12 :: Rate this Message:

Reply to Author | View Threaded | Show Only this Message

Kristian Erik Hermansen wrote:
> Dan Ritter <dsr@...> wrote:
>> ...Sun's patents suggest that they don't want anyone reverse-engineering it...
>
> You can get around the reversing-to-implementation problem using the Chinese
> Wall methodology...

A "Chinese Wall" - making sure the implementors have never seen the
original code - only helps you avoid copyright infringement. Like if
you're a PC clone manufacturer and you want to create a BIOS compatible
with IBM's.

This technique does nothing for avoiding patent infringement. That's why
software patents are so dangerous. You can do a completely independent
implementation of an algorithm, have no knowledge that the algorithm was
even thought of before, and still be in violation. Open source is
particularly vulnerable to such claims as the source is readily
available for claim holders to inspect looking for violations.

To avoid patent infringement you're better off carefully studying the
existing implementation and its patents, so you can engineer around the
patents.

  -Tom

--
Tom Metro
Venture Logic, Newton, MA, USA
"Enterprise solutions through open source."
Professional Profile: http://tmetro.venturelogic.com/

--
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.

_______________________________________________
Discuss mailing list
Discuss@...
http://lists.blu.org/mailman/listinfo/discuss