tag:blogger.com,1999:blog-4503292949532760618.post3663063202576572037..comments2024-03-16T18:42:21.178-07:00Comments on DSHR's Blog: Modeling the Economics of Long-Term StorageDavid.http://www.blogger.com/profile/14498131502038331594noreply@blogger.comBlogger10125tag:blogger.com,1999:blog-4503292949532760618.post-53695419624881122002012-02-21T09:08:57.269-08:002012-02-21T09:08:57.269-08:00Thanks to Norman Gray for pointing out that the co...Thanks to Norman Gray for pointing out that the correct link to the slides is <a href="http://www.digitalpreservation.gov/meetings/storage11.html" rel="nofollow">now this</a> and not the link in the comment above.David.https://www.blogger.com/profile/14498131502038331594noreply@blogger.comtag:blogger.com,1999:blog-4503292949532760618.post-58460206717479366882011-11-14T17:14:09.318-08:002011-11-14T17:14:09.318-08:00Wow! The effects of the floods in Thailand have be...Wow! The effects of the floods in Thailand have been severe enough that, according to <i>The Register</i>, <a href="http://www.channelregister.co.uk/2011/11/14/flash_cheaper_than_disk/" rel="nofollow">the price per GB of enterprise disk is now <i>higher</i> than enterprise SSDs</a>.David.https://www.blogger.com/profile/14498131502038331594noreply@blogger.comtag:blogger.com,1999:blog-4503292949532760618.post-8319842377892485452011-10-29T05:11:24.150-07:002011-10-29T05:11:24.150-07:00Presentations from this meeting are now on line.Presentations from this meeting are <a href="http://www.digitalpreservation.gov/news/events/other_meetings/storage11/index.html" rel="nofollow">now on line</a>.David.https://www.blogger.com/profile/14498131502038331594noreply@blogger.comtag:blogger.com,1999:blog-4503292949532760618.post-30968658779666406712011-10-27T16:12:40.929-07:002011-10-27T16:12:40.929-07:00I guess this is an awful warning against messing w...I guess this is an <a href="http://www.scientificamerican.com/article.cfm?id=finance-why-economic-models-are-always-wrong" rel="nofollow">awful warning</a> against messing with economic models.David.https://www.blogger.com/profile/14498131502038331594noreply@blogger.comtag:blogger.com,1999:blog-4503292949532760618.post-47207327286272488792011-10-17T14:25:10.773-07:002011-10-17T14:25:10.773-07:00Any true believers in Kryder's Law out there, ...Any true believers in Kryder's Law out there, for whom disk drive price drops are a law of nature, should take note of this comment from the <a href="http://www.theregister.co.uk/2011/10/17/wd_thailand_flooding/" rel="nofollow">sales director of a major disk distributor</a> (my emphasis):<br /><br />"There is no volume discount of any kind, inventories are tight and <b>prices are rising</b>,"<br /><br />The reason is that historic floods in Thailand are submerging disk factories and their suppliers.David.https://www.blogger.com/profile/14498131502038331594noreply@blogger.comtag:blogger.com,1999:blog-4503292949532760618.post-83376766442665348602011-10-11T15:56:01.000-07:002011-10-11T15:56:01.000-07:00The slides for this talk are here.The slides for this talk are <a href="http://www.digitalpreservation.gov/news/events/other_meetings/storage11/docs/27_rosenthal.pdf" rel="nofollow">here</a>.David.https://www.blogger.com/profile/14498131502038331594noreply@blogger.comtag:blogger.com,1999:blog-4503292949532760618.post-17189618218660301562011-09-30T05:51:24.348-07:002011-09-30T05:51:24.348-07:00OK, I understand the distinction.
My sense, then,...OK, I understand the distinction.<br /><br />My sense, then, is that people are not *the* major cost in *storage*. For example, it sounds like your intent is not to try to capture costs in preparing the content for ingest, and that is certainly where a lot of ICPSR's people costs reside.<br /><br />If I try to think about people costs in the narrow context of storage, then I have to ask where I draw the line. For instance, if a sysadmin-type writes a fixity checking tool and reviews the results periodically, is that preservation or storage? The person who migrates the content from storage array technology A to storage array technology B? The person who migrates the content from repository software technology A to repository software technology B? And so on.<br /><br />Maybe any storage solution of any size needs a fixed amount of people, and then scales (for some function) with the quantity of storage thereafter? And the trick for the storage manager is to select technologies and architectures where the function grows slowly?<br /><br />So perhaps the punchline is indeed to fold these into the operational and migration costs as you've suggested.<br /><br />Thanks again for the interesting model and talk.Bryan Beecherhttps://www.blogger.com/profile/18073369761415410262noreply@blogger.comtag:blogger.com,1999:blog-4503292949532760618.post-75528960134071686992011-09-29T00:49:02.161-07:002011-09-29T00:49:02.161-07:00I deliberately titled the talk "Modeling the ...I deliberately titled the talk "Modeling the Economics of Long-Term <i>Storage</i>" rather than "Modeling the Economics of Long-Term <i>Preservation</i>" because I believe that the economics of preservation are even more complex than those of storage. There's more work than I can do just in modeling storage.<br /><br />I believe that staff costs do form a major part of preservation costs. Do they form a major part of your storage costs?<br /><br />That being said, if in a particular case staff costs are significant, you could fold them into the running and migration costs in the model. The model is intended for "what if?" use, so I expect people to adjust its parameters.David.https://www.blogger.com/profile/14498131502038331594noreply@blogger.comtag:blogger.com,1999:blog-4503292949532760618.post-25297872989190855732011-09-28T11:52:17.409-07:002011-09-28T11:52:17.409-07:00David, I thought you did a pretty nice job of summ...David, I thought you did a pretty nice job of summarizing the model in the ten minute slot.<br /><br />I think a model would be a very valuable tool, and I'd like to encourage you to continue to refine it.<br /><br />As I mentioned at the LC meeting, I think it would be useful if there was some way to capture the people costs too. I know at ICPSR we spend something like $5-$10 on people for every $1 we spend on technology.<br /><br />(High-level summary of the meeting is <a href="http://techaticpsr.blogspot.com/2011/09/designing-storage-architectures-for_28.html" rel="nofollow"> here </a> if anyone is looking for additional context.)Bryan Beecherhttps://www.blogger.com/profile/18073369761415410262noreply@blogger.comtag:blogger.com,1999:blog-4503292949532760618.post-21570362240915566302011-09-27T13:21:18.987-07:002011-09-27T13:21:18.987-07:00Researchers at Purdue just announced a solid-state...Researchers at Purdue just announced a solid-state memory technology that Kryder & Kim (obviously) didn't evaluate. It is a variant of FRAM, which they did evaluate, called <a href="http://www.purdue.edu/discoverypark/global/news/news.php?id=1086&center=7" rel="nofollow">FeTRAM</a>. FRAM stores a bit in a capacitor, and has destructive read-out; FeTRAM stores a bit in a transistor and has non-destructive read-out. My amateur evaluation is that this may score better than FRAM, but probably not enough to change the overall conclusion.David.https://www.blogger.com/profile/14498131502038331594noreply@blogger.com