home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.unix.admin
- Path: sparky!uunet!pipex!doc.ic.ac.uk!cc.ic.ac.uk!imperial.ac.uk!vulture
- From: vulture@imperial.ac.uk (Thomas Sippel - Dau)
- Subject: Re: How do you back up a terabyte?
- Message-ID: <1993Jan11.145759.17449@cc.ic.ac.uk>
- Sender: vulture@carrion.cc.ic.ac.uk (Thomas Sippel - Dau)
- Nntp-Posting-Host: cscgc
- Reply-To: cmaae47@imperial.ac.uk
- Organization: Imperial College of Science, Technology and Medicine
- References: <1992Dec31.174549.677@mav.com> <1993Jan2.165300.13971@siswat.hou.tx.us> <1993Jan4.125908.28013@cc.ic.ac.uk> <1993Jan10.043714.6764@siswat.hou.tx.us>
- Date: Mon, 11 Jan 93 14:57:58 GMT
- Lines: 45
-
- In article <1993Jan10.043714.6764@siswat.hou.tx.us>, buck@siswat.hou.tx.us (Lester Buck) writes:
- - In article <1993Jan4.125908.28013@cc.ic.ac.uk> cmaae47@imperial.ac.uk writes:
- - >I fear the "perfect, reliable backup" can no longer be achieved
- .....
- - >You should give up the notion that there is such a thing as a verifyably
- - >correct filesystem backup and go for a "reliable enough backup set".
- .....
- - No, reliable backups are achievable by several methods, though you
- - may not like the price.
- -
- - 1) Applications log their transactions, so fuzzy backups can be
- - reconciled on an application by application basis.
-
- This is exactly what I implied, serious applications must be able to recover.
-
- - 2) Use a transactional filesystem. If necessary, consider the
- - data blocks and not just the file metadata as part of the
- - transaction.
-
- Which is only as reliable as the media containing the transaction log.
-
- - 3) Or go whole hog and build your filesystem on top of a database.
- - Tandom's operating system takes this approach, and the filesystem
- - is extremely reliable.
-
- "Verifyably correct" means (among other things) that the file system will
- continue to be correct and its applications can continue using it after,
- say, humanity has been wiped out by a global thermonuclear war.
-
- Thus it of similar universality as the statement "2 + 3 = 5", which will
- continue to be correct after such an event.
-
- My point was that it is futile to attempt to achieve such a level of proven
- integrity, or even to spend time arguing whether it can be achieved or
- would be useful if it was. That time is better spend improving the procedures
- intended to cope with failures (and using them!).
-
- Thomas
-
- --
- *** This is the operative statement, all previous statements are inoperative.
- * email: cmaae47 @ ic.ac.uk (Thomas Sippel - Dau) (uk.ac.ic on Janet)
- * voice: +44 71 589 5111 x4937 or 4934 (day), or +44 71 823 9497 (fax)
- * snail: Imperial College of Science, Technology and Medicine
- * The Center for Computing Services, Kensington SW7 2BX, Great Britain
-