Backup Central

Hello,

On Saturday 09 June 2007 12:08, db wrote:
Hi all

In my opinion Bacula is the best open-source backup system and one of
the best open-source projects overall. Specifically I like the
cross-platform support and easy management. However I feel that it
needs an important feature: to backup systems where just copying the
files the system use isn't enough. Systems like MySQL, MSSQL, Postgresql
and Exchange (like Symantec Backup Exec can). Backing up MySQL and
Postgresq is easy but I don't know anything about MSSQL. With
OpenChange's libmapi backing up an exchange server shouldn't be that hard.
So my question is: Are there any work going on to implement database and
exchange backup or is there at least a will to add it?
I know there are a lot of systems that need to have separate backup
modules, but isn't those 3 databases and exchange a good place to start?

The first point is that IMO, Bacula can back up all the databases that you
mention. The "problem" is that Bacula as a project has provided scripts to
backup databases that it uses. This is because we use those databases,
program them, and have looked at the documentation.

Concerning the other databases that you mention. I suspect that they all can
be backed up by techniques similar to what we have done for backing up the
Bacula catalog. It is just that we (at least to this point) have left it to
you the users, and the experts with those other databases, create your own
scripts.

There exist a number of scripts to do what you want. Some are included in the
examples directory of the source code, if they were contributed by users,
others have been posted to the email lists, and even more have been posted on
the Bacula wiki (see the web page).

Bottom line: I don't see any problem doing what you want, the tools already
exist, but as of today, the Bacula project does not explicitly maintain and
distribute those scripts/tools.

One of the projects I am personally working on is to create a wide variety of
professional services for Bacula through independent service providers. In
this context one of the things we are targetting is to collect, document, and
understand all those scripts and make them available to customers -- since
this takes a lot of effort and it is professional services, one should expect
to pay some modest fees for this. At the same time, I imagine that the
Bacula project will have access to all those scripts, or users can simply
continue to contribute them, so they will be availble one way or another
depending on whether or not you want professional support or not.

Certain commercial backup vendors have integrated solutions more directly into
their products that can sometimes be more efficient and easier for the
administrator to use. I don't foresee the Bacula project undertaking these
kinds of projects because the Bacula project is about Bacula not about MySQL,
PostgreSQL, Exchange, Oracle, DB2, ... However, it is possible that in the
context of the professional services some company or companies may want such
integrated solutions (i.e. a solution that allows admins to use rman directly
to do the backup), and they may be willing to pay for such developement. If
that happens, we will certainly encourage them to contribute any such
solution to Open Source, if possible. Despite what a sponsoring corporation
may want, many APIs are proprietary and as such probably cannot be
distributed with Open Source because of the STRICT licensing requirments of
the GPL (see my recent email on the catch-22 of GPL). Even things such as
the API for NDMP, which pretends to be an Open Standard, seem to be far from
open, and even farther from being compatible enough to be freely distributed
with Open Source.

Regards,

Kern

Kern Sibbald wrote:
I don't foresee the Bacula project undertaking these
kinds of projects because the Bacula project is about Bacula not about MySQL,
PostgreSQL, Exchange, Oracle, DB2, ...
Bacula is about Bacula? I thought Bacula was about backup? Like it or
not, but admins have to take backup of files AND systems. If Bacula
fails to acknowledge this, then it can't provide a complete backup
solution and that would be a shame.
Yes you can script a solution to some database systems, but it would be
nice to have a supported solution that was part of Bacula. Scripting a
database backup still doesn't help exchange and other systems. The
possibility to create an exchange module it here, but am I the only own
that sees the need for it? Doesn't people want an alternative to systems
like Symantec backup exec as a central complete backup system?

Best regards
db

On Saturday 09 June 2007 15:47, db wrote:
Kern Sibbald wrote:
I don't foresee the Bacula project undertaking these
kinds of projects because the Bacula project is about Bacula not about
MySQL,
PostgreSQL, Exchange, Oracle, DB2, ...
Bacula is about Bacula?

Yes, Bacula is about backing up files.

I thought Bacula was about backup? Like it or
not, but admins have to take backup of files AND systems. If Bacula
fails to acknowledge this, then it can't provide a complete backup
solution and that would be a shame.
Yes you can script a solution to some database systems, but it would be
nice to have a supported solution that was part of Bacula. Scripting a
database backup still doesn't help exchange and other systems. The
possibility to create an exchange module it here, but am I the only own
that sees the need for it? Doesn't people want an alternative to systems
like Symantec backup exec as a central complete backup system?

I believe I already explained to you how I plan to get solutions that
are "part of Bacula", and in explicitly acknowledging this issue, I cannot
undstand why you say " If Bacula fails to acknowledge this ...". At the
moment, there are *many* more critical features that Bacula lacks than what
you are asking for when one looks at the voting results from the Bacula users
themselves.

In the mean time, the users using Bacula can help the project by either
helping provide those solutions or by writing and contributing scripts.

Kern Sibbald wrote:
I believe I already explained to you how I plan to get solutions that
are "part of Bacula", and in explicitly acknowledging this issue, I cannot
undstand why you say " If Bacula fails to acknowledge this ...". At the
moment, there are *many* more critical features that Bacula lacks than what
you are asking for when one looks at the voting results from the Bacula users
themselves.

Oh my bad, I thought you meant that you had to pay for backup of
commercial systems.

Best regards
db

On Saturday 09 June 2007 15:47:01 db wrote:
Kern Sibbald wrote:
I don't foresee the Bacula project undertaking these
kinds of projects because the Bacula project is about Bacula not about
MySQL, PostgreSQL, Exchange, Oracle, DB2, ...

Bacula is about Bacula? I thought Bacula was about backup? Like it or
not, but admins have to take backup of files AND systems. If Bacula
fails to acknowledge this, then it can't provide a complete backup
solution and that would be a shame.
Yes you can script a solution to some database systems, but it would be
nice to have a supported solution that was part of Bacula. Scripting a
database backup still doesn't help exchange and other systems. The
possibility to create an exchange module it here, but am I the only own
that sees the need for it? Doesn't people want an alternative to systems
like Symantec backup exec as a central complete backup system?

What it means is that bacula's purpose isn't to provide a unified way of
backing up/restoring any software configuration you'll find, with a nice,
friendly GUI with a big 'restore database' button.

Backing up Mysql and PostgreSQL for instance is quite easy, as they are well
documented and easy to backup. But do you want to do sql dumps or binary
backups, maybe you'd even like to backup database journals during the day ?

It's funny you take Exchange as an example, as the only supported way of
backing up Exchange is, I think, NT Backup... And I'd say I trust MS on this
one, as most Exchange backup agents (symantec included) we've tested suck big
time. As it is, we backup exchange on disk with ntbackup, then through the
network with bacula. We do the same with Oracle (RMAN, then bacula).
You know what ? We already did the same before with our proprietary software,
as there were dedicated agents, but none of the worked as well (reliably) as
ntbackup or RMAN... no one knows the internals of these softwares as well as
their editors.

Anyway, what is really important, I think, is that :
- Developping a dedicated solution for one piece of software is a lot of work,
and work you have to do again for each release... even some proprietary
backup software have trouble following the pace
- Some APIs may even not be disclosed to open source, or with NDA or fees
- You'd be better off trusting the provided utility to backup each software :
pg_dump for postgres, rman for oracle, ntbackup for exchange, etc ...

Marc Cousin wrote:
It's funny you take Exchange as an example, as the only supported way of
backing up Exchange is, I think, NT Backup... And I'd say I trust MS on this
one, as most Exchange backup agents (symantec included) we've tested suck big
time. As it is, we backup exchange on disk with ntbackup, then through the
network with bacula. We do the same with Oracle (RMAN, then bacula).
You know what ? We already did the same before with our proprietary software,
as there were dedicated agents, but none of the worked as well (reliably) as
ntbackup or RMAN... no one knows the internals of these softwares as well as
their editors.

Ah ok, I see. Thank you for your reply :-)

Best regards
db