On 07/03/2005 10:55 AM, Real, Will wrote:

>Dear list,
> 
>We are about to implement a new strategy for archiving our digital image 
>masters. One copy of each file will be maintained on a live RAID server and 
>another on an off-line hard drive stored off-site. The access to the on-line 
>masters will be very restricted; nonetheless we are concerned about two 
>things: guaranteeing the authenticity of the master file, and monitoring the 
>masters for any possible degradation.
> 
>Up to now we had been using offline CDR for master image storage. We had been 
>burning the MD5 checksums of all of the images onto their CD to enable future 
>monitoring of the integrity of the file. The idea was that we would 
>periodically scan the CDs to detect any changes in the MD5 checksum, which we 
>thought would be indicative of some degradation and therefore a signal to 
>migrate the data to another medium. What we did not have was a manageable and 
>affordable procedure for monitoring the MD5 checksums for thousands of files 
>on hundreds of CDs. [We were not so worried about authenticity since we had 
>read-only files on the CD. Now however it is more of a concern because in 
>theory, though access to the files will be restricted, someone could in fact 
>replace the original file with an altered one].
> 
>Now that the files will be stored on the server, we envision some kind of 
>automated monitoring of the MD5 checksums. We picture an application running 
>in the background, or running on a schedule (nightly, weekly, etc.)  that 
>compares current checksums to the checksums originally recorded and stored in 
>a database, and reports in a log any mismatches. The same application could 
>perform this analysis on the files stored on off-line hard drives as well.
> 
>We are currently using an image browser called ThumbsPlus (version 7) which, 
>among other things, creates reference thumbnails and stores MD5 checksum 
>values in its back-end database, written in MS Access. It could be that with 
>these MD5 values stored in the image browser database, we are part-way 
>there--we have the first half of the equation. (I think the new Photoshop CS2 
>file browser, and perhaps many of the other image browsers like Cumulus or 
>Portfolio do the same thing)
> 
>Do any of you have a system of the kind we are envisioning? Is there any 
>"off-the-shelf" software that does this? Absent a pre-existing application, 
>would any of you be interested in sharing the cost of developing something 
>like this?
>  
>
It sounds like a perl script, running on schedule, could do it and that
shouldn't be expensive to develop. Perl can read and write to MDB
(Access) files, and it can do checksums. I would have it store its own
history of stored checksums, so that it's not only comparing to the
"original value"--at Brooklyn Museum, at least, there will always be
some people who'd have rights to bother replace the image, and change
the "original" stored checksum value. And you should burn the checksum
history to CD or otherwise store it in a form that's hard to alter, but
easy to access should you suspect tampering.


---
You are currently subscribed to mcn_mcn-l as: [email protected]
To unsubscribe send a blank email to 
[email protected]

Reply via email to