TSM -can I separate full from incremental backups?

[liyboviz]

Hi,



Can TSM keep different copies for full backups and incremental ?



I want to backup a client every day with incremental backup but also to backup the client a full backup every month and keep it as a different copy.

without backup sets can it be done?



backup sets - can i create backup sets for more than one node on the same tape?







Thanks

 

[cabec]

Hi.



Why don't you use two different storage pools to keep your backups seperated?



Merry X-Mas, happy new year.



Carsten.

 

[liyboviz]

Hi,



What is the configuration needed in order to backup server files daily to storage pool X and when doing a monthly backup, keep it in storage pool Y ?



Thanks

 

[LED888]

<TABLE BORDER=0 ALIGN=CENTER WIDTH=85%><TR><TD><font class="pn-sub">Quote:</font><HR></TD></TR><TR><TD><FONT class="pn-sub"><BLOCKQUOTE> I want to backup a client every day with incremental backup but also to backup the client a full backup every month and keep it as a different copy.</BLOCKQUOTE></FONT></TD></TR><TR><TD><HR></TD></TR></TABLE>



Could create two nodes.

Node A - Weekly Incremental backup

Node B - Monthly Full backup.



If we do not want to have two nodes.

Could perform a monthly archive, which is basicly a full backup with a long retention.

Also the data for the incremental is in the backuppool and the archive is in the archivepool. Two different storage pools.

If the retention for the archive pool is too long, change the retention from 365 to 180 or

90.



Some food for thougt,

Sias

 

[liyboviz]

hi,

Thanks for the archive Idea...



about the 2 nodes idea - how can I create to nodes, I can't install 2 clients on the same machine...





Thanks

 

[LED888]

You do not need to install the client code twice to get the second node.



Since I do not know what kind of operating system that the TSM Client is running on

the concept is the same with the exception of Windows.



For Unix:



In the dsm.sys file



* Stanza 1

Servername FRED

TCPServeraddress XXX.XXX.XXX.XXX

TCPPort XXXX

NODENAME Weeklybackup

.

.

.



* Stanza 2

Servername BOB

TCPServeraddress XXX.XXX.XXX.XXX

TCPPort XXXX

NODENAME Monthlybackup

.

.

.

.





There are going to be two dsm.opt file.

The dsm2.opt can be named anything we want as long as we use the -optfile parameter to

point to the file. The file could be called bob.opt, george.opt, nancy.opt or gina.opt.

I think you can even get away without using the .opt extention, as long as we have the

parameter servername that match the same parameter in the dsm2.sys file.





dsm.opt

SERVERNAME FRED



dsm2.opt

SERVERNAME BOB



For Windows, all the above information will be in the dsm.opt file since there is not a dsm.sys file.



To start the client command line:



dsmc &lt;- This will read the first stanza.



dsmc -optfile=/path/dsm2.opt



or



dsmc -servername=bob



To start the GUI for the 2nd node:

dsm -optfile=/path/dsm2.opt

or

dsm -servername=bob



Of course the nodenames need to be register with the TSM Server.

Also the TSM Server will see two nodes even though they are both on the same system,

need to register additional client nodes.



If you want to perform scheduling, need to have two schedule daemons.



dsmc sched &lt;- this will read the first stanza.





dsmc sched -optfile=/path/dsm2.opt

or

dsmc sched -servername=bob



Now with two nodes, you can have two schedule

One for weekly backup and another one for monthly backup.





Good Luck,

Sias

 

[liyboviz]

thanks a lot...



I'll see what we'll prefer in our company... 2 nodes or to use the archive option.



Thanks again.

 

[Harry_Redl]

You can use archives for your purpose ... and direct them to another pool

 

[pchapus]

I propose archive too for this purpose. But pay attention to the high cumulative volume you will get with archive. :-o



Sometime, i propose the "generate backupset" tsm function which extract all tsm client datas to one set.

 

[liyboviz]

Hi,



Backup sets - does only one node data can be written to a single tape?



I mean if I use a 400gb lto2 tape, i got space for more than one client data, is it possible to create backup sets of more than one node on one tape?



Thanks

 

[Harry_Redl]

Hi,

this is done by default (more client nodes to one tape), only if your tape pool (sequential access pool to be exact) has COLLOCATION attribute set to YES then a single node has its own private tapes (which is good for node restoration - you do not need to load so many tapes)



Does this help?

H.

 

[liyboviz]

Hi,



Yes, I use collocation, so that means that I can't create 2 backupsets on 1 tape...



Thanks

 

[Harry_Redl]

Yes, that's it ... you have to choose what you want ... if you use collocation, you know exactly which tape belongs to which node (so in the case of node failure you know you need just these tapes, not all of them - therefore restoration is faster - less mount requests)

on the other hand you need more media because you cannot mix data from more nodes on a single tape ...

 

[mrj]

Hi Sias,



I think your idea of using two nodenames is great, but don't I need to buy

two licenses per computer in that case?



Mira

----------

<TABLE BORDER=0 ALIGN=CENTER WIDTH=85%><TR><TD><font class="pn-sub">Quote:</font><HR></TD></TR><TR><TD><FONT class="pn-sub"><BLOCKQUOTE>You do not need to install the client code twice to get the second node.



(... content removed ...)



Of course the nodenames need to be register with the TSM Server.

Also the TSM Server will see two nodes even though they are both on the same system,

need to register additional client nodes.



If you want to perform scheduling, need to have two schedule daemons.



dsmc sched &lt;- this will read the first stanza.





dsmc sched -optfile=/path/dsm2.opt

or

dsmc sched -servername=bob



Now with two nodes, you can have two schedule

One for weekly backup and another one for monthly backup.





Good Luck,

Sias

</BLOCKQUOTE></FONT></TD></TR><TR><TD><HR></TD></TR></TABLE>

 

[rwhtmv]

Licensing is per processor (last time I renewed). As long as you have the license for the processors on that node, you shouldn't need any new licenses.

 

[denverdigiman]

Be aware the COLLOCATION is only used for storage pools. Since BACKUPSETS do not use storage pools, there is no concept of COLLOCATION. In fact, if you have two nodes, each that will generate a BACKUPSET of, say 25GB, and you are using LTO tapes, you will have 2 LTO tapes (one for each node) - each tape will have a 25GB BACKUPSET on it - effectively wasting 150GB of tape space



<TABLE BORDER=0 ALIGN=CENTER WIDTH=85%><TR><TD><font class="pn-sub">Quote:</font><HR></TD></TR><TR><TD><FONT class="pn-sub"><BLOCKQUOTE>Hi,

this is done by default (more client nodes to one tape), only if your tape pool (sequential access pool to be exact) has COLLOCATION attribute set to YES then a single node has its own private tapes (which is good for node restoration - you do not need to load so many tapes)



Does this help?

H.</BLOCKQUOTE></FONT></TD></TR><TR><TD><HR></TD></TR></TABLE>