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

Re: [semi-OT] Data archiving (was Re: Query on adding a USB hdd)



-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 05/24/07 08:47, Douglas Allan Tutty wrote:
> On Wed, May 23, 2007 at 10:49:51PM -0500, Ron Johnson wrote:
>> On 05/23/07 20:17, Douglas Allan Tutty wrote:
>>> On Wed, May 23, 2007 at 07:05:23PM -0500, Ron Johnson wrote:
>>>
>>>>> It would be very nice if there was a universal cross-platform rw +
>>>>> encrypt filesystem for archives.  Something that you could be confident
>>>>> that you could decrypt and access in 10 years using whatever OS was
>>>>> current then.
>>>> tar is cross-platform, as is ASCII CSV.  PGP/GPG is also cross-platform.
> 
>>> I don't know if a generic tarball I make today will be readable by
>>> whatever OS in 10 years, which is why I store a current install cd.  In
>>> 10 years, hopefully I can find a computer that will boot it.
>> You've got bigger problems if you think that a CD-R will keep it's
>> integrity for 10 years.
> 
> No.  I figure a CD is good for at least a year.  Every year, I
> pull the two netinst cds from the bank, take an SHA hash and compare it
> with the written notes, then run something like cdck on them.  So far,
> my Woody CDs are fine.  Funny enough, so is my woody floppy set (the
> whole shebang set of 20 floppies) on Maxell floppies; needed for my 486
> that doesn't boot from CD or run an installer after woody's.
> 
>> Tape (using tar, and a media used by "large data processing shops",
>> since they are supported for a LONG LONG LONG time, unlike that gee
>> whiz specialized crap that NASA seems to love) or SCSI hard drives
>> (in external enclosures so you can spin them up annually) formatted
>> ext2 or FAT32 are what I would choose.
>>
> 
> Would you use tar to make a tarball and put it on a hard drive formatted
> ext2, copy as is to ext2 (changing ctime in the process), or forgo a
> filesystem and write tar directly to the raw disk?

Copy a (possibly compressed) tarball to an ext2 volume.  That way,
you could have multiple timestamped or "different project" tarballs
on the same device.

>                                                     What tar format
> would you use: GNU or Posix?

Good question.  GNU tar does Posix.  I'd have to research the
differences between the two formats.

>> FAT has been around for 26+ years, and ext2 is 14 years old.
>>
> 
> Which is more resistant to bad blocks popping up after time in storage?  

Are either?

Maybe you'd also have to create PAR2 (forward error correction) files.

http://en.wikipedia.org/wiki/PAR2

>>> If I gpg a tarball today with whatever algorithm is current, in 10 years
>>> that algorithm may be long cracked.  Will the gpg authors keep support
>>> for it?  Perhaps.
>> This is FLOSS.  Save the source on a separate disk with SHA512 hash
>> codes.
>  
>> And text is *the* guaranteed data format.  Database backups should
>> be text format extracts and "Office" documents should be in ODF
>> format which is just zipped text.
> 
> Never heard of ODF, or is it specific to *Office programmes?
> Personally, I save my latex as latex.  The origional contents are
> plainly visible.

Never heard of ODF????  It's the OpenOffice.org 2.0 document format,
aka OASIS Open Document Format for Office Applications, ISO/IEC
26300:2006.

http://en.wikipedia.org/wiki/OpenDocument

>>> accessible?  Perhaps the time capsule would have to include a whole
>>> computer and not just the archive media.
>> Yes.  If by computer you also mean "the whole schmeer, including
>> many tape drives" since it's common on Big Systems to backup a
>> single database in parallel to multiple tape drive.
> 
> Computers keep getting smaller.  Computer could mean a little brick that
> has an interface for the archive drive, the archive drive unit, and some
> kind of user interface.  RS-232C has been around for ever; will it be
> around for evermore?  If the backup medium was hard disks, then an
> interface for the hard drive plus an enclosure for the drives if the
> brick didn't have the pysical space.

If you did that in 1990, you'd have put in an ISA controller.  Five
years ago, it would have been a PATA drive.

All these issues just go to show how difficult the subject of
digital archiving is.

- --
Ron Johnson, Jr.
Jefferson LA  USA

Give a man a fish, and he eats for a day.
Hit him with a fish, and he goes away for good!

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)

iD8DBQFGVcVTS9HxQb37XmcRArS1AJ9tthx04o3J8O/QI2X7jBSVphaWQQCdFFou
6+zW4Gjqbk/m1L3R2ZrBEAQ=
=JGee
-----END PGP SIGNATURE-----



Reply to: