Kern Sibbald writes:
When I was first researching the kind of basic organization I wanted for
Bacula, I took a look at Amanda, read a bit on their email list, and talked
to a user who had used Amanda, which basically from what I understand works
much like that. That is Amanda figures out what has to be done then does it.
I have never looked at Amanda, but Bacula reminds me a lot of a program
called Tivoli Storage Manager (TSM.. if that is still the name, it was
originaly called ADSM).
Easiest way to describe TSM to list members is Bacula in Steroids. :-)
You could move data between pools, you could control retention periods based
on location, server backed, number of files, number of files after the file
is deleted; restores were lighting fast for small data sets, you could do
point in time restores.. and much, much more.
It is a commercial program, if it still is available, and was somewhat
expensive. Thousands of dollars for a "basic" setup.
I think Bacula has a very bright future and like very much what I have seen
so far, but that doesn't mean it doesn't have lots of room for improvement.
Well, the person I talked to about Amanda and a number of other users
complained the most about precisely that feature.
In my opinion it is very difficult to make processes simple and foolproff.
Backups are significanly more complex than many people would believe.
Specially when you have to take into account bandwith of the network, hard
disks, how busy the servers getting backed up, how busy is the backup
servers, how often data changes.
For instance setting up a policy that says don't allow more than 4 hours
without a backup for data XYZ, assumes that you can backup the data within
that time period. What happens if a backup of the data runs for 6 hours.. by
then there are data that is already older than 4 hours without backup..
should you start another process?. That is just an example of the type of
problems that type of policy can have.
Bacula right now has a simmilar problem. Schedules for jobs can "pile up" if
a process takes longer than anticipated. We should have a control of some
sort that says that a particular job should just fail if a backup is still
running.
Example:
You have an incremental job that runs every 4 hours. It usually takes 30
minutes to an hour to run.
There is a particular day that lots and
lots of data will be changing for 6 to 8 hours. The periodic job that
usually takes an hour or less will take 5 hours. By the time the next job
gets created the previous one has not finished. You can set a chain of
backups that will set the server in a continued set of backups. Perhaps
after let's say 8 to 12 hours everything is back to normal, but by then you
have used lots of space. It would be nice if one could tell Bacula "start a
backup at A:B:C PM, but if a previous job is running then the job should
fail.
TSM had/has a nice way to resolve that. You could say "do not keep more than
X number of copies of a file". So if data is changing a lot during a certain
period you basically could limit how many copies you made of the data.
That said, I have often thought of adding directives to guarantee that certain
levels are performed at specified intervals
That type of automation, in my opinion, has lots of places it can go wrong.
I think may be a good project, but perhaps after more basic issues are
resolved.
-------------------------------------------------------
Using Tomcat but need to do more? Need to support web services, security?
Get stuff done quickly with pre-integrated technology to make your job easier
Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
_______________________________________________
Bacula-users mailing list
Bacula-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/bacula-users
