[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [ARSCLIST] Cost of tape vs HD (was: crashed disk drive recovery)
Happy New Year, everyone.
Dave, your points are well taken. We are obviously on the same side of
this issue (and many others). I was more worried about non-experts
(including administrators) thinking that digital (even with
high-quality drives) is cheaper than analogue.
- We must remember that we *are* asking for more money from
administrators if we follow current accepted practices (and, in
fairness, it seems fair that administrators ask for quicker access).
Again, we ask to keep the analogues *and* digital copies "forever".
- Migration as thought in the analog world had cycles measured in
decades, whereas recommended migration in the digital world happens
every six or seven years (either because of drive failure or system
updates). So in the medium-long term (despite falling prices), digital
storage appears to be quite more expensive than analog (in the
long-long term, analog may not be an option, so comparisons are
- It seems that you agree that 8€/GB/5 years may match your true
storage costs at your institution. Perhaps larger systems, despite
their greater complexity, also allow for a bit of savings through
economies of scale, or greater automation.
- Your point comparing OfficeMax drives to Certron cassettes also
raises the issue of hierarchical preservation, which is beginning to
be suggested in some circles.
We used to trust our data to oxide molecules on plastic, which
appeared to be fairly stable (with some famous exceptions), and thus
most of us did not worry too much about learning chemistry. Now we are
trusting our data to systems that from the start are potentially far
less stable if not given proper attention. As a result, we find
ourselves having to learn IT, and spend more time managing storage. It
is a new reality.
The analogue preservation world has plenty of horror stories, and the
one recounted by Richard surely will not be the last of the horror
stories from the digital world, even with supposedly "perfect"
systems. There is no question that luck also plays a part.
On Mon, Dec 22, 2008 at 3:01 PM, Dave Nolan
> Marcos -
> I never intended to say that the costs of using a managed IT approach for
> long-term storage were lower than analog - only that the current cost
> of "better quality" hard drive media was roughly comparable to the cost of
> analog reel-to-reel media - especially when when adjusted for inflation.
> Remember - my comments came out of Richard's talking about a friend/client
> with a failed hard drive, and I was trying to make the case for spending
> the few extra $$ on professional-quality drives instead of being taken in
> by the cheapest thing on the shelf at OfficeMax. Had we all done that in
> the past, our archives would all be on 39-cent Certron cassettes instead of
> reels ;-)
> And while my basic reasoning was based on media costs alone, and did not
> take into account data migration costs, neither did (almost) anyone's
> original analog "strategy" factor in the costs of preservation or migration.
> I do think that Palm's assertions in the "Digital Black Hole" article about
> admin costs are based on one particular sort of IT environment - one where
> there are indeed many many administrators for their storage systems.
> Unfortunately, the reality that I have seen - even in seemingly "large",
> well-funded institutions (a couple of prominent New York public
> broadcasting institutions come to mind) - is that instead of a well-staffed
> multimedia archiving and IT environment, there is instead a "one-person-
> shop" - or if you're lucky, a "two-person-shop" (one archivist and one
> sympathetic IT support person). This is of course despite upper
> management's perception that every piece of multimedia that has ever been
> created by the institution should be instantly available to a plethora of
> users in a wide variety of delivery formats.
> Management expectations notwithstanding, the lack of personnel and
> re$ource$ forces one to rethink the degree of in-depth management that can
> be reasonably applied by one or two people to the multimedia archive. With
> a certain amount of effort, you can automate backups and other storage
> admin duties, but in real-life practice for any library or archive that I
> have worked with, the layers of IT admin support that Palm describes in the
> Digital Black Hole article are simply out of reach - and so some levels of
> perfection are just never achieved.
> And this isn't even taking into account the well-documented problem of lack
> of multimedia knowhow and sometimes outright hostility to media archiving
> and in many institutional IT departments, which only further complicates
> the job of the digital media archivist and often forces them to become
> their own IT administrators just to build and maintain a functioning system.
> I am also noticing that if we go with Palm's estimate of just under 8 Euros
> per GB for "real" storage costs (including admin), that is not too far off
> from my storage costs for 2-channel 24/96 audio on an enterprise-class SAN
> system administered by one person. Additionally, in the three-plus years I
> have been on my current project, we have seen the price of our backup
> drives plummet by over 50%. The costs for the SAN hardware have not
> dropped quite so quickly (drives are cheaper, processors are not), but
> nonetheless we have invested in the next generation of hardware from our
> SAN manufacturer, which will allow us to house 24TB in expansion frames
> connected to a single SAN controller, as opposed to 2 TB per SAN in the
> previous generation of hardware - so our costs for primary storage are
> projected to fall, just not as quickly as with outboard backup drives.
> As for the costs of preserving the original analog (or digital, for that
> matter) media, I am perpetually amazed at the amount of money spent on what
> I consider inappropriate storage. Several places I have worked with have
> insisted on keeping their masters in a warehouse in the same city - close
> by, but with insufficient climate control, and often at exorbitant "city
> rent" costs. For some reason, there is a psychological comfort for certain
> non-archival managerial types in thinking that "geographically close"
> somehow means "more easily accessible" - despite ample evidence that the
> time and cost of retrieving a tape from the local warehouse is no less
> expensive than having it FedEx-ed from truly secure, geographically remote,
> climate-controlled storage - with overall yearly costs 1/10th of what they
> are paying to store media locally.
> Anyway - enough from me... Marcos makes some very valid points about
> presenting modern archival management as an ongoing "We've just finished
> painting the boat, time to paint the boat again..." managed IT situation
> requiring ongoing maintenance, instead of the "tapes-on-a-shelf-for-50-
> years" model that so many people have stuck in their heads when approaching
> a new project.
> dave nolan
> 92nd St. Y
> On Sun, 21 Dec 2008 14:26:25 -0500, Marcos Sueiro Bal
> <tierecords@xxxxxxxxx> wrote:
>>> One also has to consider that the costs of blank media in the recent
>>> past were at least as expensive as going with an Enterprise-class SAN
>>> system. 20 or so years ago, I was paying $10-$15 per reel of 1/4" tape
>>> holding either 30 or 60 minutes of audio. If you adjust that for
>>> ($10.00 in 1985 now equals roughly $20.00 in 2008), one can make the
>>> argument that the cost per recorded hour of 2-channel audio at 24/96 on a
>>> SAN is still cheaper than recording the same audio 24 years ago to analog
>>> 1/4" tape at 15ips (or even at 7.5ips for that matter).
>>It is true that the cost of *recording* high-quality audio is lower,
>>but for archival purposes I think these numbers are highly misleading:
>>long-term storage costs of large quantities of digital data appear to
>>incur much higher costs than analog. We are talking about five-year
>>life-cycles for the hardware/software systems (and their associated
>>expensive migrations), knowledge cost of IT (although in your
>>institution it is mostly you, you spend a lot of time maintaining and
>>researching the infrastructure), the seemingly much higher need for
>>multiple copies, etc etc.
>>Jonas Palm, in his wonderful "The Digital Black Hole", estimates that
>>the cost for *five years* of storage in his fairly sophisticated
>>institution system as €7.86/GB. Again, this does not account for
>>expensive data migration every five years.
>>Furthermore, for preservation of legacy media, the higher costs of
>>analog physical storage cannot be dismissed, since accepted practice
>>is to keep the original masters.
>>I am far from advocating for a return to analog as archival storage,
>>but for now at least (where is the 100-year hard drive?) we cannot
>>make the argument that long-term digital storage of audio is less
>>expensive than analog. I think we do a disservice to institutions and
>>clients when we present it as such. We need to know what we are all
>>Please chime in with comments, as I would love to be wrong.
>>Cheers & happy holidays,
>>Marcos Sueiro Bal
Marcos Sueiro Bal