BackupPC-users

Re: [BackupPC-users] how does backuppc handle DBs

2009-02-07 15:54:17
Subject: Re: [BackupPC-users] how does backuppc handle DBs
From: Les Mikesell <lesmikesell AT gmail DOT com>
To: "General list for user discussion, questions and support" <backuppc-users AT lists.sourceforge DOT net>
Date: Sat, 07 Feb 2009 14:52:00 -0600
Nick Smith wrote:
> Well please correct me if my thinking is not correct.
> 
> im using volume shadow, which is suppose to be a way to backup in use
> files without unmounting or corrupting them.

A shadow copy is a snapshot of the moment the snapshot is taken. You 
need to do something to ensure that the database file is consistent at 
that moment.

> the volume shadow and backups are being done after hours at a time
> when no one is logged on and making any changes to the database.

Shutting down the database before the shadow is made might work.

> so from what ive read and understand this is a safe way to backup sql
> and exchange.

It is only part of it. Once you have a consistent snapshot, shadowing 
permits the database to be live during the backup run.

> i could edit the pre/post scripts that launches the volume shadow to
> unmount the sql and exchange dbs of you guys think that would be
> a safer way.  it just adds a bit of complexity to a relatively easy
> script. and i am no programmer, i thought i was lucky to get this far.

What you are getting is approximately what you'd have if the machine 
crashed at the time you made your shadow.  If you are lucky, the apps 
can fix it when they start up.  Stopping the apps just for the time it 
takes to make the shadow will make sure the files are closed and clean. 
  Or, as someone else already mentioned, doing a dump to a more portable 
format is even better, since in a real disaster scenario you might be 
trying to restore into a somewhat different environment.

> I do understand the fact that an untested backup is no backup at all,
> i just need to figure out a good time to bring a doctors office down
> thats two hours away to do the testing.

A spare machine (or a VMware image) might be a better place to test. 
You can easily see the restored file size on any machine - and if you 
have the same database, see if it will start up and find recent entries.

> I am doing ony full backups with rsyncd over a site to site vpn
> connection.  i think what you are saying is that rsync will do a
> compare on the
> original file and only transfer whats changed correct?

Yes, it walks through the files comparing block checkums and only 
transfers differing parts.  Since database files may have their changes 
distributed within the file it is hard to predict how well this will 
work.  For something like a growing logfile it is very effective at 
finding and only sending the new part.  Also, if your vpn is doing any 
compression, you might be saving some time there since databases tend to 
have very compressible data.

 > someone else
> on here told me that repeated full backups do pretty much the same as
> an incremental with a little more cpu overhead and slightly increased
> disk usage, does that still stand as correct?

Incrementals skip over files where the length and timestamp match the 
copy in the previous full run.  Changed files (like your big DB) always 
go through the block comparison anyway.

> if it is actually only transmitting the changes in the db file, then i
> could see how the backup was only taking 40mins.  i just dont
> understand
> why it tells me im backing up 10gigs.  unless its just telling me the
> size of the file is 10gigs and its telling me its backed up.

The receiving side actually reconstructs the full file from the matching 
old blocks and the new changes, so you end up with a new 10 gig file 
which is then compressed and added to the pool.  There's no particular 
relationship between this size and the size of the new/changed blocks 
transmitted.  The one down side of the backuppc design is that large 
files with small changes between runs can't pool the large portion of 
duplicated content.

-- 
   Les Mikesell
    lesmikesell AT gmail DOT com



------------------------------------------------------------------------------
Create and Deploy Rich Internet Apps outside the browser with Adobe(R)AIR(TM)
software. With Adobe AIR, Ajax developers can use existing skills and code to
build responsive, highly engaging applications that combine the power of local
resources and data with the reach of the web. Download the Adobe AIR SDK and
Ajax docs to start building applications today-http://p.sf.net/sfu/adobe-com
_______________________________________________
BackupPC-users mailing list
BackupPC-users AT lists.sourceforge DOT net
List:    https://lists.sourceforge.net/lists/listinfo/backuppc-users
Wiki:    http://backuppc.wiki.sourceforge.net
Project: http://backuppc.sourceforge.net/