Can we achieve this separate pool configuration, or multiple
device storage in BACULA ENTERPRISE without any performance
issues.
I want maximum throughput may be for a 3000 vm's and each vm
should have their backup separated. They must be able to see
their individual backups and do what ever they want.
Dimitri,
I have heard the same complaint from you a number of times:
"Bacula sucks at disk as tapes". It is my view that your
statement is simply not true. Bacula handles tapes, disk,
fifo, cloud, aligned volumes, deduplication, and many other
device types. The original code was largely monolithic with
tests that did different things depending on the device.
However, for quite some time now Bacula has been split into
separate drivers for each device type, which means that it
is not at all correct to say that Bacula treats tapes and
disk the same. They are both handled very differently
including the device addressing -- that is each device has
very different ways of tracking device addresses and the OS
APIs that it uses to address those devices.
You can keep thinking as you wish, but it worries me that
your (what I consider incorrect) concept of Bacula drivers
may confuse other users. There are certain common factors
between each of the end devices mentioned above that Bacula
handles, but even a cursory look at the source code for
Bacula 9.0.x will prove that the details of the device are
handled very differently.
Best regards,
Kern
On 21/08/2017 19:27, Dimitri Maziuk wrote:
On 08/21/2017 11:49 AM, Phil Stracchino wrote:
I can immediately cite one practical *disadvantage*: you're going to be
doing a fandango on disk. With 500 volumes open, you may well spend
more time seeking than actually writing.
Well, I'd also want to look at the elevator scheduler and the actual
hardware too...
The problem as I see it, is bacula sucks at "disks as tapes". It doesn't
work without vchanger, and it can't auto-label volumes in the vchanger.
(Inability to write two copies at once to guard against disk failure is
icing on the cake.)
If you back up to a single filesystem, you can spool jobs in parallel to
fast drive and have them despooling to volumes sequentially to spinning
rust with very little practical speed penalty. Provided your disk
subsystem is faster than the network, of course. Adding one volume per
job/client constraint and $(JobId) to volume label shouldn't be too hard.
If you want to use removable drives as "magazines" with one drive per
client... I'd be tempted to take a close look at BackupPC or some such
before I'd commit to bacula.
------------------------------------------------------------------------------
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
_______________________________________________
Bacula-users mailing list
Bacula-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/bacula-users
------------------------------------------------------------------------------
Check out the vibrant tech community on one of the world's
most
engaging tech sites, Slashdot.org!
http://sdm.link/slashdot
_______________________________________________
Bacula-users mailing list
Bacula-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/bacula-users
