Probably as a punishment by the Fates for not
insisting of machine upgrades, I have been
unlucky enough to experience hard disk failures
on AFS fileservers.
So, I am a fan of the regular AFS volume backup
because it recovered data and "saved my life".
I don't see how a file-level backup would ever
be able to recover all the files on a dead disk.
For users, the on-line access to a mounted backup
volume is excellent and a source of wonderment.
It's so quick! No administrator help needed!
I have often pondered if the idea of backup volumes
could be extended allowing users to locate on-line
instantiations of the backup volume at several past
"vos backupsys" points instead of just the last one?
--
cheers
paul http://acm.org/~mpb
"Any technology distinguishable from magic is insufficiently advanced."
--Tom Christiansen
Shyh-Wei Luan/Almaden/IBM wrote:
> I'd like to add one point regarding to the efficiency of volume-level
> backup (butc or buta) or file-level backup with ADSM (TSM now). TSM
> file-level backup is uses a "incremental forever" model. Only changed
> files will be backed up every time (usually daily) and it does not require
> a full backup routinely. For further speeding up the file-level
> incremental backup, we have a script that examines the modification time of
> each volume and skips volumes that are not modified since the last
> incremental.
>
> I agree with Brian that a combination of file-level backup and volume-level
> backup would be better. The file-level incremental backup is good for
> quick individual file recovery (that are not in the .backup volume anymore)
> and the volume level is good for disk failure/disaster recovery.
>
> Shyh-Wei Luan
>
> "Brian T. Huntley" <[EMAIL PROTECTED]>@transarc.com on 10/03/2000 04:14:36
> AM
>
> Sent by: [EMAIL PROTECTED]
>
> To: Paul Blackburn <[EMAIL PROTECTED]>
> cc: [EMAIL PROTECTED]
> Subject: Re: AFS backup/restore using ADSM
>
> On Tue, 3 Oct 2000, Paul Blackburn wrote:
>
> // One thing puzzles me: if I run an ADSM file backup
> // (or any backup that traverses directories like tar)
> // doesn't that mean that the entire filetree being
> // backed up has to be accessed via the AFS cache?
> //
> // It doesn't seem very efficient compared with
> // doing a regular AFS backup by volume.
> // --
> // cheers
> // paul http://acm.org/~mpb
> //
>
> You are correct on both counts... everything does go through the cache,
> and it's not as efficient as doing backup by volume (for more reasons
> than just the cache). However, using a memory cache rather than a disk
> cache reduces this. We have been doing back by vol and are currently
> investigating going to a combination of back by vol and by file so that we
> don't have to keep as much data around.
>
> There are advantages and disadvantages to both methods; the trick is
> finding the right choice for your site based on data retention, backup
> window, and TSM(ADSM) server requirements.
>
> -
> Best regards,
> Brian
>
> +-----------------------------------------------------------------------+
> | Brian T. Huntley Systems and Network Engineer |
> | Campus Information Services, Clarkson University |
> | Ph/FAX: 315.268-6723/6570 |
> | [EMAIL PROTECTED] www.clarkson.edu/cis |
> +-----------------------------------------------------------------------+
> UNIX *is* user friendly. It's just selective about who its friends are.
> PGP Public Key available. finger [EMAIL PROTECTED]
