Hello,
On 3/30/2006 9:38 AM, Christoph Litauer wrote:
Hi,
I just searched the archives but didn't find hints. Sorry if this
question has been asked before.
I am using bacula-1.36.3 for the backup of about 12 linux clients. At
the moment I just use an external hardware raid as (File-)backup device:
Device {
Name = FileStorage
Media Type = File
Archive Device = /storage
LabelMedia = yes;
Random Access = Yes;
AutomaticMount = yes;
RemovableMedia = no;
AlwaysOpen = no;
Maximum Volume Size = 2147483648 # 2GB
}
The backups are done in parallel with a maximum of 20 concurrent jobs.
The most recent backup of one of the clients (about 1,6 GB of data)
consists of a full, 1 differential and 2 incrememtal backups, located on
16 backup files (FileStorage).
Running the restore I get a "Rate" of about 500 KB/sec (restoring to a
local filesystem on the storage server). The raid device is able to
deliver 12MB/sec. The local filesystem is able to write about 10MB/sec,
My conclusion for that degraded performance is that even if I use
FileStorage as a random access media, the restore process recovers the
data by sequentially reading all the 16 backup files "searching for the
right files". I would have expected a rate of nearly native disk speed
because the backup files could be "seeked".
As far as I know, there are some problems using seeking instead of
reading and interpreting which lead to Kern disabling volume seeking for
file volumes, so your conclusion is right.
A solution would be to not perform concurrent backups. But then the
backup window gets significantly wider (especially for the incrementals).
Any solutions, hints? Would my problem disappear if I upgrade to 1.38.x?
Perhaps, I don't know if that peculiar problem is resolved by now. You
might read through the release notes of the 1.38 versions.
Possible solution, assuming you've got enough disk space: Use spooling.
It's - theoretically - nonsense doing this with file volumes, but might
reslove the performance problems you see. Try to put the spool space
onto different physical disks and a different filesystem than the real
storage space!
Arno
Thanks a lot in advance. If you need more of my configuration files, no
problem, please tell me.
--
IT-Service Lehmann[EMAIL PROTECTED]
Arno Lehmann http://www.its-lehmann.de
---
This SF.Net email is sponsored by xPML, a groundbreaking scripting language
that extends applications into web and mobile media. Attend the live webcast
and join the prime developer group breaking into this new coding territory!
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=110944&bid=241720&dat=121642
___
Bacula-users mailing list
Bacula-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/bacula-users