[Pasig-discuss] Write-blocking a RAID

Tue Nov 6 04:49:20 EST 2012

The elephant in the room - and the whole reason that you sometimes need
to preserve servers rather than bits (at least in the first instance) is
that it is highly likely to be a database driven application. Thus there
is a likely to lot of implicit knowledge in the application code without
which the data files on disk are largely meaningless. Short of some
miracle there will be pretty minimal documentation for this code so the
only way to comprehend it is to actually have it running. Then you can
backtrack through the code, extract the implicit knowledge and start to
do something about extracting an repurposing the data in a more
tractable and preservable way. File formats are the least of your
problems here. My experience is that the amount of material that arrives
like this is non-trivial and likely to grow.

The original notion of write-blocking RAID's is a non-starter anyway.
Any reasonable RAID system will write to the array members on startup
even if it just basic synchronisation information. Write blocking them
will cause the controller hardware/software to consider the devices
failed, with unpredictable results. The standards for forensic
non-repudiation are much higher than is realistic or cost-effective for
archival purposes (despite what archivists think!). What you can do is
image the RAID disks individually (and then reassemble the RAID array
for the next bit) so you can restore the RAID array to its original
condition by rewriting the disks.

What you actually want is to image the filesystem as presented by the
RAID system to the OS. Traditional disk imaging tools like dd are fine
for this with the server running normally. This image can then be
mounted in a virtual machine and either booted/migrated as-is or,
probably better, mounted read-only on a fresh installation of the OS in
a VM and then have the code copied across to produce an operational VM.
This can then be snapshotted to give you a known starting point for
subsequent operations. If anything goes wrong you can restore the RAID
members and start again (at least until one of the disks fails).

This could be a case where preserving the bits, and worrying about file
formats, will only get you so far. It's about understanding and
capturing the context in which files were created and used. A bit like
books and letters, really.

Neil        

From: pasig-discuss-bounces at asis.org
[mailto:pasig-discuss-bounces at asis.org] On Behalf Of Michael Seadle
Sent: 05 November 2012 19:19
To: pasig-discuss at mail.asis.org
Subject: Re: [Pasig-discuss] Write-blocking a RAID

I agree with David and Thomas. Integrity (keeping the bits safe) is the
first priority, and I think knowing what the bits genuinely represent
(authenticity) also comes high. 

Format changes are a very manageable if the bits are there.

Best wishes ... Michael

On 11/5/2012 7:32 PM, Youkel, Thomas wrote: 

I truly have to agree with David on this one. Keeping the bits safe is
the priority

Thomas Youkel
Chief, Enterprise Systems Engineering
The Library of Congress
202-707-8355 - desk
202-615-7794 - mobile

-----Original Message-----
From: pasig-discuss-bounces at asis.org
[mailto:pasig-discuss-bounces at asis.org] On Behalf Of David Rosenthal
Sent: Monday, November 05, 2012 1:16 PM
To: pasig-discuss at mail.asis.org
Subject: Re: [Pasig-discuss] Write-blocking a RAID

On 11/05/2012 10:05 AM, Mark Fitzsimmons wrote:

	Hi Cory,

	Great point. You are absolutely right, file formats will become
obsolescent.

	This will happen over time to particular formats that you have
copies of. 

	You will inevitably need to convert/preserve the content of
those 
	files to an accessible media type.

	An inventory of the files you have and by type will provide the 
	reference point you need to track the impending migration of
them 
	ahead of obsolescence.

Whether or not at some point in the future there is a risk of format
obsolescence, the files need to be extracted and put some place
reasonably safe, as soon as possible. At this stage, worrying about
formats is a waste of time and effort. Worry about formats after you
have all the bits safe.

It is truly amazing how each and every discussion of digital
preservation gets hijacked by the "OMG format obsolescence" meme.

See http://blog.dshr.org/2012/10/formats-through-time.html

        David.

----
To subscribe, unsubscribe, or modify your subscription, please visit
http://mail.asis.org/mailman/listinfo/pasig-discuss
_______
Save the Date - PASIG Dublin, October 17-19, 2012
_______________________________________________
Pasig-discuss mailing list
Pasig-discuss at mail.asis.org
http://mail.asis.org/mailman/listinfo/pasig-discuss

----
To subscribe, unsubscribe, or modify your subscription, please visit 
http://mail.asis.org/mailman/listinfo/pasig-discuss
_______
Save the Date - PASIG Dublin, October 17-19, 2012
_______________________________________________
Pasig-discuss mailing list
Pasig-discuss at mail.asis.org
http://mail.asis.org/mailman/listinfo/pasig-discuss

-- 
Dean, Faculty of Humanities / Dekan, Philosophische Fakultät I
Professor & Director, Berlin School of Library and Information Science 
(Institut für Bibliotheks- und Informationswissenschaft)
Humboldt-Universität zu Berlin
Editor, Library Hi Tech
Blog: Digital+Research=Blog <http://digitalplusresearch.blogspot.com/> 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.asis.org/pipermail/pasig-discuss/attachments/20121106/511f3364/attachment-0001.html 