Index Retention

[databoy26]

Our retention policy has been 1 month for as long as I have been doing backups. We recently were told to do a once a month full and keep it for 10 years. I am not sure if I set this up correctly and am looking for some advice.

before I start.. We are networker 7.2.1 on Solaris 8. 90% of our backups are NDMP from a NetApp. The fulls in questions only pertain to the NetApp backups.

I created new NDMPfull and NDMPfullIndex pools for these backups. I have a full scheduled to run every 1st day of the month and skip every other day. This backup has a 10 year retention. I then have a normal NDMP pool and Index pool, which does an incremental on the 2nd day of the month, and then levels 2,3,4,5,6,7,2,3 etc... for the rest of the month.

I have this all set up now, and I am testing it, but I am finding that my Index tapes for the 10 year retention have only a 30 retention. The actually data for the full show 10 years, but the index is only 30 days. I cant seem to find where I can modify the retention of the Index tapes.

I have the browse and retention policy set for 10 years for these backups. I know that it is not advisable, but it is what I was asked to do.

thanks,

 

[hkuhlmann]

In my opinion you do not need to store your index any longer than a few days. Every time you backup your index, it holds the information for your 10 year savesets.

 

[databoy26]

I thought that it would hold the index data for 10 years.. but the volume that it writes it to is set to expire in 30 days. I thought since the backup has a 10 year retention, then the index would too.

mminfo -a -r 'volume(7), location(10), pool(7), %used,space,volretent' |grep -i ndmpfull

FUL001L2 p7000 NDMPFullIndex 4% 06/22/06
FUL002L2 p7000 ndmpFULL 33% 05/22/16
FUL003L2 p7000 ndmpFULL 0% undef

 

[605]

I think you are just cought by the fact that the index is a saveset of the NW server's client and not of your filer's client. As a consequence, index backups follow the policies for the NW server's client. This was pure NW logic so far.

However, as it is hard to convince the customer abbout this fact, this has changed with NW 7.3. The index backup policies are now the same as for the client data save sets.

 

[databoy26]

Thanks 605 - This makes more sense now.. from a logical standpoint i guess i really dont need any index on tape anyhow since my browse and retention policy are both 10 years.

One more question... From a best practices standpoint, what is the preferred time to do keep your browse policy.. I know have to have my retention set for 10 years.. but my management wants to be able to browse these for 10 years as well. If I did shorten the browse policy to say 1 year, how would I do a restore of something older than 1 year. Would I need to use scanner then?

Life was much easier when our retention was only 30 days.

 

[605]

Both policies are totally independent ... as long as you do not try to make the retention shorter than the browse policy ;-)

The key point is the retention policy. As long as it has not expired and - as a consequence of the other save set's policies on this media, the volume has become recylable AND NetWorker has not relabled the media (only if it needs one) - you still have your save set information and you can still recover.

The file index is a 'nice-to-have' but it is not really necessary. Worst case, you can either rebuild the CFI for the save set (scanner) or you restore the whole save set to a temporary location and get your files from there.

The browse policy is just a residual from the times where you did not have TB of hard disk available and the index database had limited features and capacity. Today it is 18 Exabytes per client. Enough for your purpose?