[nylug-talk] CD-R as archival media

[Steven Kreuzer]

To be honest, I think you are being a little paranoid. CD-R's have a 
lifespan of about 10 years. From a cost standpoint, CD-R's are going to 
be the cheapest. Burn the data you want to archive, put it in a jewel 
case and store it somewhere at room temp and you should be fine. If you 
are really paranoid, simply make 2 copies of the data, just in case you 
accidently run over one of the discs with your chair (this happened to 
me, don't laugh).

-Steve

On Friday, May 2, 2003, at 12:31 PM, David Panofsky wrote:

> Ok, so I have some data which I don't need frequent or quick access 
> to, but
> which I'll almost certainly need to use in the future.  I want to 
> archive it
> so I can free up some space on my system.  I've been planning on just 
> burning
> the files to a couple of CD-Rs, but I'm concerned about the 
> possibility of
> data loss due to media degradation.  I'm looking for an efficient 
> method to
> minimize my risks, and I don't really care if the data is stored in a 
> standard
> or user friendly way.  Note, I'm looking for technological advice, not
> information about off-site or fire-proof storage, etc.
>
> First off, I'm wondering if any of you have an opinion of how likely I 
> am to
> have any problems down the line.  I know that the red-book format has 
> some ECC
> built in which helps with small scratches, but I'm sure we've all 
> experienced
> cases where that is not enough protection. I have the sense that CD-Rs 
> are
> pretty likely to develop significant errors, but if I'm being overly 
> paranoid,
> just let me know....
>
> The simplest thing is probably to make multiple copies of each disk on 
> the
> assumption that no two will fail in exactly the same way. This can get
> cumbersome with a lot of data, but I'll do it if nothing better comes 
> up.  I
> was also thinking I could use an old, small (by today's standards) HD 
> to hold
> one data copy and CD-Rs for other copies on the assumption that only 
> one of
> the two media types is likely to fail in a given storage environment.
>
> Here are some ideas I've been having which use some sort of Forward 
> Error
> Correction scheme to add redundancy to the data set, but none is ideal:
>
> 1) For each 700MB CD-R, make an (X) MB ISO file and use something like 
> RAS
> (http://www.cleaton.net/ras) to create a (700-X) MB file which could 
> be used
> to recreate the original ISO assuming errors are somewhat limited.  
> Both files
> would be burnt onto the CD as separate sessions.
>
> 2) Same as above, but run algorithm on tgz file and make ISO out of 
> tgz, the
> generated redundancy information, and the software/instructions for 
> recovery.
>
> 3) Use same method with the complete set of standard ISO images to 
> create a
> single "redundancy CD" which could be used to recreate any damaged CD 
> image
> from the rest of the set.  The RAS software was actually written for 
> this
> purpose, but using floppies. This method would be a pain because of 
> the large
> temporary storage spaces involved to create/use the redundancy 
> information,
> but unlike the other options would survive a TOC failure of one of the 
> disks.
>
> 4) Make a tar file of the data and then use rar 
> (http://www.rarlab.com) to
> compress, split, and add redundancy to the file.  Each segment then 
> goes on a
> CD.  My main concern here is that while rar is a cool archive format 
> for
> including redundancy options, it is not free software.
>
> Anyway, I've already wasted enough of your time, but would love any 
> advice
> regarding this matter which you might have. Possibly there is already 
> software
> to do this sort of thing, or maybe a filesystem / archive format which 
> has
> error correction which I could burn onto the CDs.
>
>   Thanks in advance,
>   dave p.
>
> __________________________________
> Do you Yahoo!?
> The New Yahoo! Search - Faster. Easier. Bingo.
> http://search.yahoo.com
> _______________________________________________
> The nylug-talk mailing list is at nylug-talk at nylug.org
> To subscribe or unsubscribe: 
> http://herzl.nylug.org/mailman/listinfo/nylug-talk
>