--+HP7ph2BbKc20aGI
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable

* adam AT tsatech DOT com <adam AT tsatech DOT com> [2003-11-08 21:12]:
> Hi all,
>=20
> I am new to Netbackup and I thought I understood exactly how multiple dat=
a=20
> streams worked.
>=20
> I have a new policy with (2) clients and the file list include /home, /ap=
ps,=20
> and /u01.  No multiplexing and Multiple data streams is enabled.  The max=
 jobs=20
> per client has been chaged to (4) and the max concurrent drives per backu=
p on=20
> the STU is set to (2) since I have a 2/20 library.
>=20
> File List
> NEW_STREAM
> /home
> NEW_STREAM
> /apps
> NEW_STREAM
> /u01
>=20
> Yes, /home, /apps and /u01 are all on their own separate disks.
>=20
> When the policy executed the Activity Monitor shows (6) jobs with (2) bei=
ng=20
> active for client1. Example (policy ran for the first time on newly label=
ed=20
> tapes).
>=20
> 6 BACKUP QUEUE .... CLIENT2 ....
> 5 BACKUP QUEUE .... CLIENT2 ....
> 4 BACKUP QUEUE .... CLIENT2 ....
> 3 BACKUP QUEUE .... CLIENT1 ....
> 2 BACKUP ACTIVE .... CLIENT1 ....
> 1 BACKUP ACTIVE .... CLIENT1 ....
>=20
> However when I look at DETAILS for JOBID 1, it says it is mounting tape A=
00008=20
> to write /home, and when I look at details for JOBID 2, it says it is mou=
nting=20
> tape A00007 to write /apps.  Then it seems to grab either tape A00007 or=
=20
> A00008 to backup /u01 (which ever becomes available first).  Is this the=
=20
> normal behavior?  I thought that a policy like this would only use one dr=
ive=20
> and put everything to a single tape.=20

Nope.

It's doing what it is supposed to. When you multistream, it creates a
new job for each stream, which then behave as independent jobs. This
way, you can get multiple tapes spinning for a single job and reduce
your backup time.=20

> Does using multiple data streams necessarily means backing up to all avai=
lable=20
> drives at the same time?  It seems to me sort odd since this behavior wou=
ld=20
> mean that parts of the backups are split up to different tapes and to res=
tore=20
> would require more than one tape. =20

Yep. That's its job. It is NetBackup, after all; not NetRestore ;-)
The idea is that you are usually trying to reduce the time it takes to
back up data when you multistream. In general, if you have a robot, it
shouldn't really matter that much _what_ tape has your data, as long as
NetBackup knows where to _find_ the data.

Do you have a specific issue you are trying to address by
multistreaming? Or are you just trying to understand what multistreaming
is doing? If you are trying to solve a problem, let us know and maybe we
can help you find the solution.

--=20
David Rock
david AT graniteweb DOT com

--+HP7ph2BbKc20aGI
Content-Type: application/pgp-signature
Content-Disposition: inline

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.7 (GNU/Linux)

iD8DBQE/sEYEMrO4/Yb/xwYRAg9WAJ9WV66CWwCPWLKaZNOGp87CxRO+0ACgy2aC
HoYrtrOKN+tHeJq0fD0s29o=
=jCjn
-----END PGP SIGNATURE-----

--+HP7ph2BbKc20aGI--