Hello,
On 2/21/2006 9:30 PM, Wolfgang Denk wrote:
In message <[EMAIL PROTECTED]> you wrote:
C) You run multiple jobs in parallel and want to keep jobs together on
tape (to allow much faster restores, usually).
How exactly is this working, especially when the spool size is
smaller then each of the backups?
Let's say I have 3 jobs "A", "B"" and "C" which all use the same
volume and run in paralel; they all write more data to tape than the
spool device can hold.
I understand all three of them will start simultaneous, filling their
respective spool files. Assume "A" is fastest, and fills it's spool
file up to the maximum size.
* I guess the SD now starts despooling the spooled data of "A", while
jobs "B" and "C" continue to run? The FD for job "A" is blocked
now? Is this interpretation correct?
Yes, as far as I know.
* Assume now "B" fills it's spool file.
* I guess that once despooling for "A" has completed, two things will
happen: (1) the FD for "A" continues to spool new data, and the SD
starts despooling the data of "B". Is this assumption correct?
Also as far as I know and can observe.
In the end, we will have a tape / some tapes, where the data of the 3
jobs are interleaved in big blocks of the size of the respective
spool files. Is this correct?
Yes.
If yes, then how will a restore be much faster compared to a tape
where all "A" data are consecutive, followed by all "B" and then all
"C" data?
Because the comparison is not what I had in mind. You compare to having
run the three jobs sequentially. If you run the jobs simultaneously,
without spooling (which I, by the way, never tried) the data is
interleaved on tape, probably on a block level, like ABCABCABCABC...
where each block would be the the usual 63kB Bacula writes by default.
In case of a restore, Bacula would read the whole run of blocks but only
need 1/3 of it.
Preferrably, and achieved with spooling, is having larger blocks: After
reading one of the larger blocks, Bacula could (hopefully, depending on
the tape drive capabilities, the actual amount of data written in one
run, and the size of one tape file (space betwen EOFs) reposition the
tape to the next block and wouldn't have to spend time reading data it
doesn't need.
Hope this explains it a bit,
Arno
Best regards,
Wolfgang Denk
--
IT-Service Lehmann [EMAIL PROTECTED]
Arno Lehmann http://www.its-lehmann.de
-------------------------------------------------------
This SF.net email is sponsored by: Splunk Inc. Do you grep through log files
for problems? Stop! Download the new AJAX search engine that makes
searching your log files as easy as surfing the web. DOWNLOAD SPLUNK!
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=103432&bid=230486&dat=121642
_______________________________________________
Bacula-users mailing list
Bacula-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/bacula-users
