Re: [kern@sibbald.com: Re: [Bacula-devel] Re: future large programming jobs]

[Kern S. <ke...@si...>]

Hello Marc,

On Thu, 2004-05-27 at 17:52, Marc Schoechlin wrote:
> Hello Developers !
>=20
> In February 2004 I exchanged some mails with Kern about a potential
> "last semester project" for me - and you made some very interesting
> proposals for enhancing bacula for me. I started now this project
> two weeks ago - my employer defined such a bacula-related project for
> me - so I hadn=B4t so many possibilities to make my own decision :-))
> (unluckily)
>=20
> I=B4m (my company too) very interested to support bacula with the
> results of this project, so maybe some of the things we develop
> can be published as opensource or provided to the bacula-project.
>=20
> Maybe this would be interesting for the bacula-project.=20
>=20
> My current conceptual formulation of this project can be described
> like that:
>=20
> LF.net, a local ISP in Stuttgart/Germany (my employer), intends to use
> bacula as their backup-system for their data-center. We are using
> open-source software now for over 10 years, in almost any e(>90%) Envir=
onment
> of our company.=20
>=20
> Currently we have about 200 servers and 32 workstations which are
> backuped with amanda - but we are not very happy with this, because it
> has many architectural problems.=20
>=20
> We would like to use a opensource backup-software there are some
> important reasons:
>=20
>   * Purchasing a commercial backup-software is relatively
>     expensive - and this would rise our prices for hosting
>=20
>   * we are using many different platforms i.e
>     FreeBSD, 2-3 linux-distributions, SunOS,...
>     (getting actual clients for all of these platforms
>      is not very easy)
>=20
>   * our backup-server(s) should run with freebsd
>  =20
>   * we would like to interface internal management and=20
>     accounting-systems with the backup-software
>=20
>   * we would like to be able to fix software-problems=20
>     by our own
>=20
>   * ...
>=20
> My job for the next three months would be the following:
>=20
>   * evaluating bacula and identifying problems which prevent
>     usage in a large-scale production-environment.

>     (especially scalability questions....)

I'll be particularly interesting in your findings on this question.

>=20
>   * thinking about strategies to setup the=20
>     backup-environment
>=20
>   * writing software which extracts accounting-data from=20
>     the bacula-database to an other database which stores
>     accounting-data.
>    =20
>     i.e.
>    =20
>      - amount of backup-data
>        our prices for system-backups are currently not
>        dependent on the backed amount of data
>       =20
>      - prediction of future environments
>        we would like to have better analysis about our
>        backups=20
>        (i.e. when it=B4s time to buy new hardware)
>    =20
>    * writing a module for our accounting-system
>      (it=B4s a self-development of our company)
>=20
>    * writing a interface which makes it more easy for=20
>      our support-staff to fix backup-problems
>=20
>       - monitoring of the finished backups
>       - discovering of changing file-system structures
>         to maintain the filesets
>       - ...

This work should be of a lot of interest to many commercial sites
especially ISPs. Billing is a critical component of any commercial
operation. I'd be pleased if you would carefully check the data that is
going into the database to ensure that it is accurate. I've checked it a
number of times, but the whole process is reasonable complicated that
there is certainly room for error.

>=20
>    * writing a interface for our customers which will be integrated
>      in our CRM-interface
>    =20
>       -> getting information about the status of finished and=20
>          planned backup-jobs
>       -> getting information about the amount of backuped
>          data
>       -> ...
>=20
>=20
> So what am I doing currently ?
>=20
> I set up three machines which run one director and two
> storage-daemons(one LTO-1 drive, and obe file-storage). I=B4m currently
> backing up 32 office-machines with an absolute amount of a full save
> of 450 gigabytes. All of these machines are running with freebsd=20
> 4.7 and higher (the bacula-dir and bacula-sd are running with
> a partially patched freebsd4.9, because of the pthread-bug of
> freebsd).
>=20
> I=B4m testing bacula now for two weeks, and I think is is really
> good - some things need a workaround or some improvements for the
> software, but I think bacula is a really great piece of software!

Don't hesitate to mention what workarounds or changes are necessary on
the list. It is possible I am planning to implement them or some user
has already done so.

>=20
> Do you have suggestions ? :-)

At the moment I have no specific suggestions other than to say that I
think it is an interesting and valuable project, and I look forward to
hear your results as you develop them.

Best regards, Kern