ecosys
asked on
Incremental backup without a full
We are using NTbackup with along with tapes to perform our daily backups. The tapes are rotated on a weekly basis and a new tape is inserted every Tuesday.
On Tuesday nights, NTbackup performs a full backup of the directory. From Wed-Mon NTbackup perform incremental backups of the same directory.
The NTBackup's full and the incremental backups of the directory both backs up to the same .bkf on the tape.
At the end of the week, when the tape is to be rotated, the .bkf is near double the size of the directory it has been backing up. Can someone help me see what is going wrong?
I'm thinking maybe I don't need to execute a full backup job on Tuesdays and just run the incremental all week. Does anyone know if the incremental backup would perform a
a full back does not already exist on the tape.
ASKER CERTIFIED SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
I'll say again:
No, the only job type that does a full backup is a full backup. Incrementals do incrementals. Differentials do differentials.
Also:
In short, I'm not sure why you're doing an Incremental... you should be doing Fulls and differentials. Incrementals, except in rare circumstances on Windows, don't make much sense.
No, the only job type that does a full backup is a full backup. Incrementals do incrementals. Differentials do differentials.
Also:
In short, I'm not sure why you're doing an Incremental... you should be doing Fulls and differentials. Incrementals, except in rare circumstances on Windows, don't make much sense.
ASKER
Hi Leew,
Thanks again, sorry if it seemed like I was repeating my question to you. My intensions were to correct my typo so that forthcoming reader would understand my published question.
Thanks again, sorry if it seemed like I was repeating my question to you. My intensions were to correct my typo so that forthcoming reader would understand my published question.
First of all, I agree completely with leew. Diffs are 90+% better (i.e. less headache on the 'restorer') than an incremental.
A full has to precede either though.
In short, (probably full detail at the link leew posted),
A full backup goes through all files, and resets a bit on each file that's copied.
The difference comes in, with the subsequent job(s).
A differential will backup all files that have had that bit turned back on (meaning (a) the file is new or (b) the file has changed). But a diff will not reset that bit - so it will backup the same data the following night, plus any additional that has the bit altered.
The incremental will do the same, but it resets the bit. So data backed up the day after the full is ignored the following day. So, in truth your incrementals should not necessarily be growing in size. However, I know zilch about your environment and how heavy your data is hit by your users. It is possible that 10% of the data is altered the first incremental day, then 10% more is altered and another 5% added, etc... until you see the results that you are seeing. The bkf figures you posted still do not cause great concern for me.
On the restore side, the further you get from your full backup, the more tapes will be needed for an incremental.
For a diff, all you need is the diff recorded on the date you lost the data and possibly the full. 2 tapes vs potentially 7...
A full has to precede either though.
In short, (probably full detail at the link leew posted),
A full backup goes through all files, and resets a bit on each file that's copied.
The difference comes in, with the subsequent job(s).
A differential will backup all files that have had that bit turned back on (meaning (a) the file is new or (b) the file has changed). But a diff will not reset that bit - so it will backup the same data the following night, plus any additional that has the bit altered.
The incremental will do the same, but it resets the bit. So data backed up the day after the full is ignored the following day. So, in truth your incrementals should not necessarily be growing in size. However, I know zilch about your environment and how heavy your data is hit by your users. It is possible that 10% of the data is altered the first incremental day, then 10% more is altered and another 5% added, etc... until you see the results that you are seeing. The bkf figures you posted still do not cause great concern for me.
On the restore side, the further you get from your full backup, the more tapes will be needed for an incremental.
For a diff, all you need is the diff recorded on the date you lost the data and possibly the full. 2 tapes vs potentially 7...
ASKER
Thanks Leew and Sirbounty, both of you have been very helpful.
ASKER
> Does anyone know if the
> incremental backup would perform a a full back does not already exist
> on the tape.
Sorry, here is what I meant to say: Does anyone know if the incremental backup job would perform a full back first, if a full back up does not previously exist.
I'll take a look at the link.
SirBounty, The first backup (full backup) of the directory is 52 792 416K and after 6 incremental backup jobs added to the .bkf the total size for the .bkf file is around 85 000 000k