|
Re: [BackupPC-users] Block-level rsync-like hashing dd?
2011-04-12 22:38:00
Chris Parsons <Chris.Parsons AT petrosys.com DOT au> wrote
on 04/12/2011 10:12:59 PM:
> On 13/04/2011 11:21 AM, Timothy J Massey wrote:
>> Remember, you have to get them to another
machine, so ZFS or any
>> other snapshot by itself won't finish the job: you then need to
copy
>> them somewhere else. This includes SAN based solutions (though
in
>> that case, "somewhere else" might just be defined as
a different
>> storage shelf...).
> Ah, but that is where ZFS still provides the
solution. Getting the
> data somewhere else is as simple as:
> [root@solaris]$ zfs snapshot master/data@1
> [root@solaris]$ zfs send master/data@1 | zfs receive slave/data
>
> The slave server, will now have an identical copy of the master.
> (Make it readonly for now however)
But *very* likely not any more efficiently than using
LVM, possibly with parted.
> And the next time,
> [root@solaris]$ zfs snapshot master/data@2
> [root@solaris]$ zfs send -i master/data@1 master/data@2 | zfs
> receive slave/data
>
> You will have only sent the differences between this snapshot and
> the last across.
I reserve judgements until I actually see this done,
but if it's done using a log-based structure instead of a block- (or extent-)
based structure I can certainly see the benefits here.
> So unlike ext3 + snapshots, ZFS will allow you
to efficiently
> transfer your pool to another server.
Yes, very interesting. Kind of like BackupPC
itself, the first transfer will not be terribly fast, but subsequent transfers
could potentially be much faster.
Has anyone actually done this in production with realistic
amounts of data (and data turnover) to see how it *actually* works? Remember,
for example, that in theory LVM snapshotting is also the greatest thing
since sliced bread, too, but in practice there are actual rough edges to
be aware of.
>> So, I stick by my 4 options, with one of
them possibly slightly renamed.
> You might have me by definition on this. However,
I like to think of
> ZFS as significantly more capable than other snapshot techniques.
Sure, but it's still the exact same technique. I
am fully willing to accept that ZFS may offer facilities that make it a
better choice than LVM, but it's still the same principle. I would
*really* like to see that in action...
> However, it comes with the obvious downside of having to run
> Solaris.
Which for me, to date, has been a *real* downside.
I've tried to give both OpenSolaris and Nexenta a fair evaluation
within the last 6 months, but both were poor contenders, especially in
hardware support: anything beyond commodity hardware was a tough
fit. (No, I have no real interest in Sun hardware...)
That and I have too many scars from using Sun userspace
utilities for too long... :)
Timothy J. Massey
------------------------------------------------------------------------------
Forrester Wave Report - Recovery time is now measured in hours and minutes
not days. Key insights are discussed in the 2010 Forrester Wave Report as
part of an in-depth evaluation of disaster recovery service providers.
Forrester found the best-in-class provider in terms of services and vision.
Read this report now! http://p.sf.net/sfu/ibm-webcastpromo _______________________________________________
BackupPC-users mailing list
BackupPC-users AT lists.sourceforge DOT net
List: https://lists.sourceforge.net/lists/listinfo/backuppc-users
Wiki: http://backuppc.wiki.sourceforge.net
Project: http://backuppc.sourceforge.net/
|
|
|
