[prev in list] [next in list] [prev in thread] [next in thread] 

List:       bacula-devel
Subject:    Re: [Bacula-devel] Movement detection and funding
From:       Kern Sibbald <kern () sibbald ! com>
Date:       2008-02-23 13:57:06
Message-ID: 200802231457.06883.kern () sibbald ! com
[Download RAW message or body]

On Saturday 23 February 2008 13.20:56 John Enok Vollestad wrote:
> Kern Sibbald wrote:
> > Hello,
> >
> > On Saturday 23 February 2008 08.02:45 John Enok Vollestad wrote:
> >> The following is a showstopper for what I want to use Bacula for at
> >> work:
> >>
> >>     Item: 1 Accurate restoration of renamed/deleted files
> >>
> >> Since comercial alternatives is far from cheap it might be less
> >> expensive to fund the development of this features in Bacula.
> >>
> >> Do anyone have any estimate of what this would cost and how long it
> >> would take to get ready for production?
> >
> > This project is already well underway, so no funding is required, though
> > a really nice contribution to the Bacula project or the developer would
> > be much appreciated.
> >
> :-)
>
> Will send a portion of what we save.

Nice.

>
> Is it possible to get a receipe of "yearly / qarterly support" or
> something like that - preferable as a PDF with logo so it does not look
> like charity but more like paying for work - which is how i view it.

Well, we don't yet supply paid for support (coming soon), but we can "bill" 
for development work, feature implementation, or customization. I'll explain 
the possibilites a bit more in detail offline.

>
> > The code is implemented and in what I would call a pre-alpha state. It
> > currently exists as a patch and if my review of the code goes well and on
> > schedule, the code should be integrated into the SVN trunk by next
> > Monday.
> >
> > There will then be some more work (certain details are not yet complete),
> > and a *lot* of testing to do.
> >
> > Since we are all volunteers doing implementation for free, we do not feel
> > bound by any fixed schedules.  However, that said, I am currently
> > planning the official production release of the next version (which will
> > have Accurate backup) by the end of June 2008.  It can be tested at any
> > time, but should be ready for beta testing (of that feature) within about
> > a month.
>
> We will follow the email-lists and start testing it then.
>
> > A good amount of the very low level routines of Bacula have been
> > re-written from scratch in this version, so the needed testing goes far
> > beyond just testing particular features.
> >
> >> Understood that there might be
> >> an opportunity to use the validation mechanism during an backup.
> >
> > I don't understand what you mean in the above sentence. Could you
> > explain?
>
> Probably my misunderstanding / forgetfullness.  I do not find the source
> now . Have tried to read a lot the past days and I thought that it was
> mentioned in a bug-report that one could use mechanisms used in
> validation when implementing this project but I can not find it now so
> probably just my eyes that crossed.
>
> >> We can of course do a lot of the testing.
> >
> > That would *really* be appreciated.
>
> Is there any lists of tests that should be run?

We have a "regress" portion of the SVN that has regression scripts that every 
serious user should be using to verify Bacula with their hardware.  It is 
documented in the README in that directory as well as in the Developers 
guide.  The accurate restore regression test is not yet part of the normal 
tests since it isn't yet committed, but it will be.

>
> >> We are currently using Legato to tape (LTO2) today but the restore time
> >> is too long and it lack movement detection as far as I have understood.
> >> Backup to disk would therefore require a lot of full backups to have
> >> backups without too large differences from the source thus require a lot
> >> more diskspace thereby making it more expensive.
> >
> > Well, I would really be pleased if we can provide an important feature
> > that Networker does not have.  I do know that TSM has this feature, but
> > as with Bacula the admin must explicitly enable it as it requires
> > additional resources (searches) to implement.
>
> The detection of moving files and directories is not implemented on
> Legato to my knowledge.  It is implemented on Networker and is called
> "move detection".
>
> Legato were installed here 3,5 years ago partly because of the plugin
> support.  This plugin support is now less needed as the services that
> needed special plugins is now inside virtual computers that we backup
> with snapshots and then copy as ordinary files.

OK, I probably misunderstood. I didn't realize there was a Legato.  To me, 
Legato was the name of the company before being bought by EMC and Networker 
was the product (or one of them) that Legato made, but then I really don't 
pay much attention to those companies :-)

>
> >> We have the following system to backup:
> >> 10TB of data
> >> 55GB of changes every day
> >>
> >> Retention time 1 year but we do not need to keep every single day for a
> >> full year.
> >
> > That sounds pretty typical (not particularly large) of the kinds of
> > Backups that are being done using Bacula.
> >
> >> As for hardware we wanted to use an areca 1680 controller and
> >> SAS-extenders between some supermicro boxes and large disks.
> >
> > Uh, hopefully someone else can comment on those since I don't know what
> > they are (well supermicro boxes and big disks do make sense) ...
>
> For those interested:
>
> One nice thing with the supermicro boxes is the ability to daicy-chain
> with SAS-extenders.  The Areca controller also have the opportunity to
> online increase the size of raid-sets and volumes in combination with
> raid 6, global hotspare, up to 128 disks and good performance.  Combined
> we can add shelfs and disks as needed.
>
> One may put a motherboard in or one can use it just as a shelf.  There
> are contacts for the SAS-extenders in the back of the box.
> http://www.supermicro.com/products/chassis/3U/836/SC836E2-R800V.cfm
>
> The Areca controller:
> http://www.areca.com.tw/products/pcietosas1680series.htm
> http://tweakers.net/benchdb/testcombo/1628
>
> http://tweakers.net/benchdb/test/114
> http://tweakers.net/benchdb/test/100
>
> One of the things I like most with the Areca controllers is that the
> disks in a raid-set contain the raid-information and can be moved to a
> newer or other type of areca-controller or another newer disk-shelf.
> One can splitt the disks using two shelfs - as long as the controller
> see all the disks needed for the volume it should work.  It can also
> spin down disks not used.  And management is possible out of band by an
> ethernet port on the card.

Interesting, thanks.  Well beyond my budget.

Best regards,

Kern

>
> > Best regards,
> >
> > Kern
>
> Thanks a lot :-)
>
> John Enok
>
> > PS: I have added bacula-users to the list so that they can see what is
> > coming :-)
>
> Hope my reply fits both lists.



-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2008.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
_______________________________________________
Bacula-devel mailing list
Bacula-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/bacula-devel
[prev in list] [next in list] [prev in thread] [next in thread]