[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Recommended Linux Backup

> On Thursday October 29 2009 5:14:44 pm lrhorer wrote:
>> 1. Back up to removable hard drives
>> 2. Span multiple target volumes
>> 3. Maintain a virtual fileysystem so all snapshots look like a single
>> backup to the user.
>> 4. Maintain an easily monitored index so the user can see which drive
>> will be needed for a particular backup or restore operation.
>> 5. Be able to easily rebuild the index and virtual file system from
>> the backup drives (preferebly just one drive) if the database is lost
>> on the source system.
> tar will do 1, 2, and 4. It might be able to do 5.

        Tar will *NOT* do 1 & 2.  If tar reaches the nd of a tape volume, it
prompts for another volume (it knows how to handle an EOT).  When
writing to a file, if the need of the volume is reached, tar aborts. 
Since tar does not maintain an index at all, it cannot do either 4 or

> How will you reconcile requirement 3 with requirement 2? That is, if
> you have multiple target volumes that may not all be mounted, how will
> the user to able to see and interact with his or her backups? If you

        Many archive utilities create indices of their archives which are
stored on media other than the backup media (or in addition to it). 
Enterprise level software like Amanda and Bacula use a database like
MySQL to maintain their indices.  Others use a simple file.  When
dealing with backups for hundreds, or even thousands of hosts, a
database approach is essential.  In much smaller systems like mine, a
simple file based set of indices is fine.

        There is nothing new or unusual about any of my requests.  I was just
looking for some experienced advice.

Reply to: