Re: [ARSCLIST] Archiving to hard drive

[Pip Laurenson <Pip.Laurenson@xxxxxxxxxxx>]

Does anyone have any thoughts regarding the pros and cons of the
different data tape formats? AIT, DLT, LTO?

-----Original Message-----
From: Joel Bresler [mailto:joel.br@xxxxxxxxxxx] 
Sent: 05 March 2003 03:19
To: ARSCLIST@xxxxxxxxxxxx
Subject: Re: [ARSCLIST] Archiving to hard drive


Sue wrote:

>You're likely to have to migrate off the cd's or server at some point 
>in the near future...

Thanks for the helpful post. Once the music is on a server (presumably
on a hard disk) why would it necessarily need to be migrated? SCSI
drives have built in error correction, and the data could be mirrored on
another hard disk for security. Isn't this the most likely long-term
data storage solution, bar none?

Best,

Joel

At 02:47 PM 3/4/2003 -0700, you wrote:
>  Responding to Claire Michelle Viola's request for information:
>
><While we have been transferring here and there to CD-R, mainly in 
>order to provide service copies, the audio engineers here at Eastman 
>are now seriously considering the option of transferring directly to 
>hard drive, with several servers to provide for back-up.>
>
>
>The Naropa Audio Archive Project is in a one-year pilot (NEA 
>funded)project to test the migration of analog audio cassette from 30 
>years of lectures, performances and panel discussions onto various 
>digital formats, including hard drive, for preservation and for access.

>We've been moving over audio since October.
>
>Both considerations - preservation, and access - have specific 
>requirements that need to be addressed before any material is migrated,

>or you run the risk of having to redigitize it.
>
>For preservation - neither cd-rs or hard drive are 'approved' 
>preservation mediums for audio at this time. That means your DAT is 
>your oficial preservation copy.  Any format you migrate to is 
>provisional and needs to be as pristine a copy as you can afford 
>(uncompressed file, high sampling rate, professional equipment).  
>You're likely to have to migrate off the cd's or server at some point 
>in the near future, so budgeting for multiple (continual) migration is 
>a consideration. As is the format and encoding you intend to employ in 
>creating your digital copy of the audio.
>
>For access, how are you going to catalog the transferred files, how do 
>you want access to be handled from the server, what kind of 
>permissions, copyright, or releases are necessary, how are you going to

>log material as it's transferred for which cataloging format...gotta 
>work all that out first.
>
>We have a small audio lab equipped with a professional grade (nakamichi
>dragon) cassette playback, (for your project this would be replaced by 
>DAT playback machine). After converting analog to digital the signal is

>transferred in realtime into a digital audio workstation (computer with

>audio programming - ours is a SADiE). We're sampling at the highest 
>rate we can afford to store on - 44/24. We want a high-quality 
>preservation master, because the cassettes are going to be unplayable 
>soon - even if we are calling it our 'preservation' copy. Is it worth 
>it to go even higher? Everyone we talk to disagrees.
>
>Once the file is in the computer, we create:
>-one CD-DA (audio)for reference - this is a flat, unaltered transfer 
>-two CD-Rs in BWAV format - also the unaltered file (BWAV is the most 
>open format we could find - we want the files to be able to be read in 
>20 or 200 years). We burn copies of the audio files on two different 
>manufacturers' CDs because there seems to be almost 10% bad cd stock on

>a random basis. We're using Mitsui gold and Taiyo Uden, although I 
>don't think TU is making gold alloy anymore. Anyone know another 
>supplier? This is a preservation issue - gold alloy being most stable.

>We may move over to DVD's as standards emerge. -one back up BWAV file 
>on AIT data storage tape in case the CDs reach some as-yet-unknown end 
>of lifespan
>
>As we transfer into the computer, we log the material for all fields 
>required in Dublin Core cataloging. We'll test capturing info 
>sufficient for Marc/Ead records next month - then we'll make 
>test-catalog records for both to see what level of cataloging we can 
>accomodate on a moving-forward basis.  We want the public and 
>researchers to know the material's available, so cataloging for digital

>web-based access is very important to us. We're also logging the 
>content so as to better describe the files. We're not creating 
>transcripts at this time.
>
>All copies made to this point go to a temperature and humidity 
>controlled storage facility as the preservation copies. We intend to 
>check the cds and AIT tape annually for loss of data and migrate as 
>needed.
>
>-For access, we take the original file that's still in the SADiE, and 
>put it through noise and hiss reduction programs. We edit out the 
>intermissions, smoking breaks, and other dead air. We create an access 
>protection/dubmaster CD-DA (audio copy) and a listening copy.
>
>This is at a 3-1 ratio - three hours of real time to process one hour 
>of original audio. It takes up a lot of CDs. At 44/24 every hour of 
>original uses up 3 cds for each BWAV copy. Our librarian does not want 
>to store and handle the 15000 or so access cds we expect to make. So 
>we're developing plan to test server for digital access to collection.
>
>We will still create CD-rs and AIT tape for preservation. There's too 
>much fragility to count on keeping preservation copy on hard drive, at 
>least for us. We're in a flood plain. Too much memory for us to afford 
>uncompressed hard drive copies of the entire collection. We'll rig up 
>output from SADiE into server to move compressed access files over. 
>Server will have RAID array as well as a tape back up system. We're 
>working on the front-end internet design now. We figure 2.5 tb to store

>3000 hours of compressed audio.
>
>The compressed MPEG files will go out to folks over the internet after 
>the internet user registers online and signs use agreement. We are 
>planning to offer streaming audio as opposed to downloadable audio, but

>we'll have to see what's most useful to people.  What we are worried 
>about is how many people might use it at once - I have no idea how to 
>budget for bandwidth until we're up; still looking at what to do if too

>many simultaneous users attempt to listen. We should be so successful.
>
>Hope this is helpful.  If you'd like more specifics, budget info,etc, 
>don't hesitate to give me a call. I were you, I'd call Seubert and 
>Seadle to advise on the project design. They have been very helpful to 
>our project.
>
>Sue Salinger
>Administrative Director
>Naropa Audio Archive
>2130 Arapahoe
>Boulder, CO 80302
>(303)546-3573
>archive@xxxxxxxxxx



Joel Bresler
250 E. Emerson Rd.
Lexington, MA 02420
USA

781-862-4104 (Telephone & FAX)
joel.br@xxxxxxxxxxx