Video Screencast Help
Symantec to Separate Into Two Focused, Industry-Leading Technology Companies. Learn more.

EV DB Backup - Full VS differential

Created: 22 May 2013 • Updated: 21 Jun 2013 | 8 comments
Sarah.Seftel's picture
This issue has been solved. See solution.

A customer of mine asked me why should he care for the DB backup alerts showing in console.

Meaning:

He has a very large DB, so he cannot back them up fully on daily basis.

So, he backup the DB once a week fully, and in the rest of the days he performs differential backups.

He wants to move to Full backup every two weeks and continue with differential backups.

His question to me is "why symantec monitors the Full backup state"? and wants to know if there can be any issues to have full + differentials, or is is a EV system must to have full backups.

Any ideas on why system only monitors Full state? and not seeing that the system was backed in different way?

Sarah

Operating Systems:

Comments 8 CommentsJump to latest comment

Rob.Wilcox's picture

My guess is that it's down to the rate of churn in the database.  Differentials/incrementals are all well and good if data isn't changing too much, but for things like the Vault Store database and so on, each archiving run there is (I imagine) a lot of change to the database, as well as new data.

JesusWept3's picture

Out of all the EV Datasets, including indexes, vault store partitions etc, the databases are the smallest things to backup, and them being large single files, they're quicker and easier to backup than say small DVS files or the index files.

Typically a lot of orgs will do a backup of the database to disk and then backup from that BAK file that SQL Server created, then after a few days delete the files from disk

That way, if something happens to the DB, you dont have to pull from tape, you have it there on disk, and if anything happens to the disk itself, you can pull from the tape etc, makes things a lot quicker

But honestly, the db backups are something I would not mess around with, some losses can be unrecoverable, and some could put you out of action for weeks running EVSVR and Index Rebuilds

For instance i know that if you lose all your databases, and restore from say, one week ago, but you dont restore indexes, then the indexes will fail because the Internal Index Entry has a higher Index Sequence Number than what the database has, and if that index fails, you can't run EVSVR to re insert the items

So you have to rebuild the index and then run EVSVR , and EVSVR isnt the quickest of processes etc

Advisor's picture

In addition, you can have look at following technote: (Its for EV and Backup Exec though it clarifies your query of Full backup)

http://www.symantec.com/docs/TECH194919

I hope this helps..

And below technote describes how EV determines the last date of DB backup, just FYI.

http://www.symantec.com/docs/TECH195072

Sarah.Seftel's picture
Hi
 
The first tech note says:
 
This is by design Incremental backups cannot be used to restore the databases, only full backups can be used to restore database in the event of a disaster recovery
This is not completely correct, as differential backup joined with full backup and logs can be used to return from a disaster recovery.
 
It have checked with several DBA's that I know, and they all approved that from SQL 2008 R2, you can have weekly full backup and daily differentials & transaction logs.
This should be updated by Symantec in the article.
 
However, my question is - why would EV care in which method I backup and alert on backups that are not full? 
Or to put is in other words:
As long as I have Full + Diff + TRN Logs, I can ignore those errors?
Sarah.
 
AndrewB's picture

if you know the backups are setup and occuring successfully, why not change the EV monitoring settings to suit the environment?

Andy Becker | Authorized Symantec Consultant | Trace3 | Symantec National Partner | www.trace3.com

Sarah.Seftel's picture

I am going to change the monitoring according to my backups, BUT - I'm trying to understand what is behind the request to back up the DB full and why is EV system monitoring only full backups.

This question was raised by a customer, to my openion - there was probabelly a reason to request full backups on daily basis, I would like to understand what the reason is, and is there a different recomendation from Symantec since SQL changed majorly in the last years, but the backup best recommendations stayed the same.

AndrewB's picture

when i look at the monitoring tab (site settings) i see the option to configure monitoring for full and incremental. imho, backups and level of backups are going to be based on the comfort level of the customer and, of course, the backup technology they have. i've seen everything that tony and JW mention and even customers that set the dbs to simple recovery mode so that they have to do full backups each day. bottom line, if you lose your dbs you're in for a world of hurt.

p.s. just as a matter of interest, how big are the databases in this particular environment?

Andy Becker | Authorized Symantec Consultant | Trace3 | Symantec National Partner | www.trace3.com

JesusWept3's picture

Full backups are bullet proof, however requesting restore job after restore job of fulls, incrementals, differentials, replaying transaction logs, can mean longer down time, lots of question marks over data integrity etc
And I've seen a lot of times when people think they have different backups on tape but they don't etc

Anywho, I could be wrong but I don't think differentials and incrementals trigger that backup flags in SQL to be reset and it's those details logged in the msdb's that EV reads last backup off of

SOLUTION