Archiving Digital History at the NARA

val1s writes "This article illustrates how difficult archiving is vs. just 'backing up' data. From the 38 million email messages created by the Clinton administration to proprietary data sets created by NASA, the National Archives and Records Administration is expecting to have as much a 347 petabytes to deal with by 2022. Are we destined for a "digital dark age"?"

Comments

16000 formats?!?

[gardyloo]

Hm. This sounds like a job for OpenOffice...

347 petabytes?

[ravenspear]

Ok, I was tempted to make a pr0n joke about this, but I think the bigger question is what kind of indexing system will this use?

I haven't seen any software system that can reliably scale to that level and still make any kind of sense for someone that wants to find a piece of data in that haystack, err. haybarn.

Re:347 petabytes?

[OrangeSpyderMan]

I haven't seen any software system that can reliably scale to that level and still make any kind of sense for someone that wants to find a piece of data in that haystack,

Haven't you? Have you ever worked with real archiving before? IBM have some nice solutions that allow us to stock on disk and a WORM library (Tivoli Storage Manager) and index in a (large) Oracle DB - they work and scale just fine (our experience over a couple of hundred teras). You probably wouldn't want all that data in a single archive anyway, but i'd guess you'd know that if you'd ever archived anything....

Re:347 petabytes?

[ravenspear]

Well considering that Spotlight took about 2 hours to index my 120 GB drive, that would be (347 * 1024^2) * 2 = 72771174 hours = 83,000 years to index that much data.

Now I'm sure the gov would use a faster system than my laptop, but still!

Re:347 petabytes?

[CodeBuster]

The most common structure used to index large amounts of data stored on magnetic or other large capacity media is the B-Tree and its variants. The article linked here explains the basic idea of the balanced multiway tree or B-Tree. The advantage of this type of index is that the index can be stored entirely on the collection of tapes, cartridges, disks or whatever else while only the portion of the tree which currently being operated on need be read into volatile or main memory. The B-Tree allows for efficient access to massive amounts of data while minimizing disk reads and writes. Theoretically, the B-Tree and its variants could be scaled up to address an unlimited amount data in logarithmic time.

Data loss will always be a possibility

[divide overflow]

It happened with the Great Library of Alexandria, with pagan libraries throughout the Christian era, and more recently has happened with antiquities in Afghanistan and Iraq. The only thing that can reliably preserve data is large scale, geographically widespread distribution of copies.

Re:Data loss will always be a possibility

[tabdelgawad]

Actually, it's more like 'inevitable'. I'll bet almost everyone has unintentionally lost digital data permanently and will do so again in the future.

The key, I think, is prioritization. We all do it individually (important stuff gets backed up many times and often, unimportant stuff perhaps never backed up), and NARA will have to do it too. I don't think backing up a president's email and backing up some minor whitehouse aide's email should have equal importance. The trick will be to come up with a reasonable prioritization scheme that will make the probability of losing the most important stuff very small.

Re:Data loss will always be a possibility

[writermike]

It happened with the Great Library of Alexandria, with pagan libraries throughout the Christian era, and more recently has happened with antiquities in Afghanistan and Iraq. The only thing that can reliably preserve data is large scale, geographically widespread distribution of copies.

True. But I hardly think Alexandria was lost to the tap of the Y key, a pregnant pause, then an "oops."

Re:Data loss will always be a possibility

[Tristor]

No, but it could have been lost to the strike of flint, a pregnant pause, then an "glukús theométôr" (Sweet Mother of God, for you people that suck). (Note: I spent like 20 minutes transliterating that to Latin just so I could post it on /. because it hated the Greek charset. I have no life.)

Answer is Compression?

[reporter]

National Archives and Records Administration is expecting to have as much a 347 petabytes to deal with by 2022. Are we destined for a "digital dark age"?"

Perhaps, the answer is compression.

Does anyone know whether there is an upper limit to text compression?

In signal processing, there is a limit called the Shannon Capacity theorem, which gives the maximum amount of information that can be transmitted on a channel. In text compression, is there a similar limit?

Note that the Shannon Capacity theorem does not tell you how to reach that limit. The theorem merely tells you what the limit is. For decades, we knew that maximum limit on a normal telephone twisted pair is about 56,000 bits per second, according to the theorem. However, we did not know how to reach it until Trellis coding was discovered, according to an electronic communications colleague at the institute where I work.

If we can calculate a similar limit for text compression, then we can know whether further research to find better text compression algorithms would be potentially fruitful. If we are already at the limit, then we should spend the money on finding denser storage media.

Re:Answer is Compression?

[MasterC]

The only thing that comes to mind is information entropy. If you're given a text document, you can determine the probability distribution for each letter, letter combinations, for words, or whatever you can think of. Then given the probability distribution, you can determine the information entropy. If, in the sum, you use log with base 2 then H(x) (see formal definitions) gives you the entropy in bits.

For example, if you have a text file with letters of equal probability (all letters have a probability of 1/27) then the bits required to represent a single letter turns out to be ~4.7549 bits. (Indeed, 2^4.7549 = 27)

This is the upper limit of compression. Such methods as the, now 50-years old, Huffman coding do decent work at approaching this limit (used in JPEG, for one).

So the answer to your question is: it's not broadly definiable for "text" or "information" but based on the patterns of the English language or a specific document.

Re:Answer is Compression?

[zysus]

Actually there is an upper limit...
It is some of Shannon's work on Information Theory.
Basically, information has entropy associated with it. Entropy being the randomness of information. Truly 100% random information cannot be compressed.
The central idea has to do with the probability of something occuring.
Text compresses quite well because certain letters are more common than others and there are a limited number of symbols. (e for example)
If i encode e using 1 bit instead of 8 that saves 7 bits.

This is the idea behind Huffman Coding.

Binary data... well, depends on the data.
I ran into this at work... basically, I was trying to reformat some data to save space on disk and eventually figured out that bzip would accomplish the same thing.

ha

[The Big Ugly]

"Archiving Digital History at the NARA"

You'll have to pry it from my cold, dead hands!

Ohhhh, NARA, not NRA....

Retain it all.

[d3m057h3n35]

Perhaps it would be best to keep it all, even the stuff that now may seem totally useless, like Clinton administration emails from Janet Reno to Madeleine Albright asking what she thinks about Norman Mineta and his "hot Asian vibe." With search technology improving constantly, it would probably be better than throwing stuff away which could potentially be of interest, or spending time developing the AI to make the task less time-consuming. And besides, we can't make future historians' jobs too easy. They've gotta earn their pay, reminding us of the banalities of this age.

Google to the rescue!!!

[feloneous cat]

With the new GoogleNARA...

nara.google.com

Oh, wait... I'm getting ahead of myself...

Difference between data and trash

[HermanAB]

In the age of pen and paper, only important stuff was written down. Nowadays all crap is preserved. This is useless. There is a big difference between data and information.

Dark Ages

[TimeTraveler1884]

Are we destined for a "digital dark age"?"

If by "dark age" you mean a time in human history where more information is recorded than ever, yes I suppose we are.

I think more accurately, we are headed towards an age of super-saturation of information. I have no doubt we can store all the data we are currently and will be generating. The question is how do we process it in to something meaningful? Just because we have the ability to archive everything, does not mean it will be useful to the [insert personally welcomed overlord] of the future.

Maybe historians of the future will be fascinated that Clinton's instant-message signoff was "l8ter d00d", but I doubt it. We'll want to save everything now of course, because we can. But the majority of the information I suspect will just be filtered out when actually searched.

Personally, I take the "you never know" ideology and save everything.

Not a dark age... was the past so bright?

[G4from128k]

Digital technologies mean that archivists now enjoy orders of magnitude more information than they had in the past. Consider all the hallway and phone conversations or jotted notes lost in a paper-based organization versus having an archives of e-mail, IM, and sticky-note digital files.

Digital technologies mean that archivists now enjoy orders of magnitude more potential accessibility that in the past. Even if paper has greater innate archival lifespan, its physical form makes in inaccessible to all but a select monkish class of archivists colocated with their paper archives. Even the select few archivists who are allowed access to paper archives can only effectively process at best dozen documents per minute (and only a dozen per hour if they must wander the files to find randomly dispersed documents).

By contrast, digital technologies radically expand access on two dimensions. First, technology expands the number of people that can access an archive in terms of distance -- a remote researcher can have full access, including access to documents in use by other archivists. A low cost to copy documents means a wealth of information. Second, search tools provide prodigious access to the files -- searching/accessng/reading thousands or millions of documents per second.

To say we face a dark age is to presume that paper documents provided far more enlightenment and comprehensiveness of documentation than paper ever actually did.

Answer is not compression, it's less data.

[gus goose]

People should think outside the box.

The answer to archiving the required volumes is producing less volumes. Case in point... we recently spent a week or so at work optimising a process that was I/O bound. The bugger took 10 hours to run. Although purchasing faster disks, converting to RAID0, and other techniques did whittle down the execution time to about 5 hours, the final solution was to redefine the process to reduce the actual IO (removed a COBOL sorting stage in the process), and the process is now 2 hours.

Bottom line: with the 100 + 38 million dollars (FTFA) assigned to the project I am sure I could eliminate a number of redundant positions, optimise some communication channels, retire voluminous individuals, replace inefficient protocols/people, and basically reduce the sources of data. Hell, if the US were to actually have peace instead of demand it, there would be a much reduced need for military inteligence, political rhetoric, and other civil responsibilities. The military could be half the size, and what do you know, we could not only reduce the requirement for archiving, but could actually save money in the process.

Remeber, govenment is a self-supporting process.

Go ahead, mark me a troll.

gus

Re:Answer is not compression, it's less data.

[MasterC]

...other techniques did whittle down the execution time to about 5 hours, the final solution ...is now 2 hours.

That's only a 60% reduction. A 60% reduction of 347 PB is still 138.8 PB...still a huge archival task.

Keep 1% of the data still leaves you with 3.47 PB. Not impossible, but still a daunting task.

burn, knowledge, burn

[Leontes]

The ancient, esteemedgreat library of alexandria was burned to the ground as knowledge literally turned to smoke, lost to mankind forever. Was it barbarians? Motivated by political revenge? Demanded by religious zealots? Accidental byproduct of an act of war?

Really, it's only the great works of artistry that need to be retained and remained, sustained and maintained. Historically, it's interesting to catalogue art, but politics? The everyday communications that lead up to the horrible decisions that lead our politicians to make the mistake of the daily business? We want records of this?

Perhaps the easiest way of keeping this knowledge at all interesting or inspiring is to burn it regularly, let people imagine what happened to allow such blunders or let apologists spin tales of delight explaining elegant solutions to how stupid people stumbled upon genius decisions. Conspiracy theorists or intellectual artistry can probably generate far greater truths than the truth will ever reveal.

It would save a great deal of money too, just having a delete key. If we are going to care so little for the decisions in the here and now, why preserve the information to be twisted by people in the future with their own biases and projects? We seem to care so little for truth knowadays, why should that change in the future?

Re:burn, knowledge, burn

[mrogers]

Doesn't it diminish the aura of a great work of art if you know that it can always be restored from a backup?

Re:burn, knowledge, burn

[mcrbids]

Really, it's only the great works of artistry that need to be retained and remained, sustained and maintained. Historically, it's interesting to catalogue art, but politics? The everyday communications that lead up to the horrible decisions that lead our politicians to make the mistake of the daily business? We want records of this?

Absolutely, yes!

History is often taught as "Charlamagne took over Constantinople in the year 12xx" as though military feats really mattered to the average Joe. But, the truth is, America was colonized by people who thought that, however bad it might be in a virgin land, it was BETTER than their lives in Europe.

One of the key failures in public education today is to communicate the understanding that history is comprised mostly of PEOPLE doing ORDINARY things in their time to make life better for themselves and their families. They loved, worked, got bored, and cracked jokes at the expense of their leaders, just like we do today.

History doesn't consist of battles, anymore than history consists of artworks. Capturing more detail in the average, everyday lives of people gives a much better understanding to the cultural norms, and the ideals to which people aspired.

The pyramids of ancient Egypt provide a clear, artistic monument to their culture, yet we have an only modest understanding of their day to day cultures. Similarly, we have Stonehenge as a clear monument to the grooved-ware people of the English isles, but almost NO understanding of who they were and what they felt was important. How much would a true historian give to understand the day-to-day culture of these mysterious "grooved-ware" people of ancient?

Those memos and IMs comprise that understand of people today.

Re:Usually when I archive...

[ArchAngel21x]

I guess you didn't see how Mr. Ebbers or the founder of Aldephia are facing prison time. Quit trying to spread that liberal lie that white collar crime pays off. By the way, it is inappropiate to refer to blacks as niggers. Grow up and learn to be a little more tolerant of diversity.

So?

[ArchAngel21x]

By the time the government comes up with a half ass solution, archive.org will already have it all organized, online, indexed, and backed up.

Have a look at the Fedora Project

[pangloss]

http://www.fedora.info/
(Not to be confused with the Linux distribution)

From the website, Fedora is "a general purpose repository service...devoted to...providing open-source repository software that can serve as the foundation for many types of information management systems".

Problem for some is that Fedora can be a little hard to grok. It's not an out-of-the-box repository to install and run, like the repository application mentioned in the article (DSpace). It's an architecture for building repository software. Once you understand the potential for building applications on top of Fedora, you start to see some light at the end of the tunnel for just the sort of issues the article raises.

Re:Dark Ages are ahead! All aboard

[screwthemoderators]

I think it may be worse than that- that there will be a huge proliferation of false information, sensationalistic 'infotainmnet,' advertising, propaganda, etc... Why, historians of the future may be depending on /. as their main source of of information! Think of what a tragedy that would be!

Relevant, interesting post

[Council]

Here is a relevant post by Ralph Spoilsport on an earlier article, which can be found here. I am reproducing it here in full because it is very interesting and highly relevant.

this is actually a BIG question

And one that I have railed about for many years.
I have been in the same position the Author discussed, and I have come to ONLY negative conclusions. In a few words, and I hate to say this, but buddy:

WE'RE FUCKED.

Digital is a loser's proposition. backing up to analogue or even digital data on analogic substrates (such as DV tape) fail. Simply nad purely.

The *only* thing that comes close is some kind of RAID, and those, even with the plummeting price of storage, are still too expensive given the needs.

Also, a RAID assumes a continuity of several things that are not likely to be continuous:

With Video:
Framerate, number of lines, colour depth, aspect ratio, file format, compression format, Operating system compatibility, etc etc etc. All of these things are variables.

With Audio:
sample rate, compression format, bit depth, file format, etc.

Basically all of it points to very bad places.

I am fairly well convinced that our age will simply disappear. They will find our garbage, the few books not pressed on acidic paper, our paintings (fat lot of good the abstract stuff will mean to them) and drawings, that's about it. the rest will just be shiny little bits of crap in the landfill.

Since we will have used up all the dense energy forms, they will be appalled at the energy requirements just to get the few remaining museum piece devices to work. Archiving the 21st century will be impossible. To the 25th century, the 21st century will be seen as a dark age - not only for the holocaust of the die caused by the failure of the petroleum based economy, but from the simple fact that very little of the information formats we are totally geared into will survive, including this note on /.

His problem of saving personal video is just the tip ofthe iceberg. His problem is the problem of our very civilisation, writ small.

That's why I am abandoning video, and going back to painting. In 500 years, my painting CAN survive. the video simply won't.

RS

And don't give me shit about my karma or whatever. My karma's fine, I don't care about it. I'm copying this because it's interesting and contributes to the discussion.

What do you think about Ralph's thoughts?

Re:Relevant, interesting post

[fmaxwell]

I think that he's being absurdly pessimistic. Those future historians will have no more difficulty reading our media than we have playing the sounds from a wax cylinder for an Edison phonograph. Running archived computer software will be no different than any of us using a software emulator to run a game for a long-dead gaming console.

Sure, optical and magnetic media decay, but there's nothing stopping people from "refreshing" the media before it decays too far. If you have a stack of CD-R discs that are starting to show an increase in correctable errors, then you back them up to new CD-R discs or to DVD-R. You don't have to sit idly by and watch them decay. I've got a CP/M computer with that's over 20 years old and it can still boot from its 10MB (yes, megabyte) hard drive. So it's not like data just disappears five years after it's recorded.

It's also a problem which is being addressed by the industry. There are companies offering long-life CD-R media designed for archival. Other companies offer data storage for archival data, much like the climate-controlled vaults where countless audio master tapes and films have been stored for decades.

In closing, I think that 95%+ of archived data will still be able to be accessed in a century -- provided that it is properly stored and cared for.

Slightly overdramatic?

[mrogers]

Are we currently experiencing a dark age because we don't have access to every letter, memo, bank statement and laundry ticket created in the 20th century? Archiving everything is an attractively simple approach, but if it turns out to be impractical we can always fall back on common sense and restrict ourselves to archiving the maybe 10% of things that have even a remote chance of being interesting in 100 years' time.

Tanks for the Memories

[Doc Ruby]

We need to imprint holographic storage on synthetic diamonds. Even if they're slow and expensive, they'll last even longer than the paper records they replace. We'll have to spend a fortune redigitizing all the polymer (CD/DVD, floppy, tape), celluloid (microfilm/fiche) and rotating (disc) media that will age to illegibility within our lifetimes. Until we get holographic gems, we need to archive everything on paper, including those expiring media, in a format easily digitized to a more permanent medium. But of course the government, and barely unaccountable bosses, want the public record to disappear down the memory hole. If they could accelerate the process, including newspapers, they'd spend everything we've got (and more) to make it happen.

Records

[Big Sean O]

NARA makes a distinction between a document and a record. Any old piece of paper or email is a document, but a record is something which shows how the US government did business.

For example, the email to my supervisor asking when I can take a week's vacation isn't a record. The leave request form I get him to sign is a record. An email about lunch plans: not a record. An email to a coworker about a grant application probably is.

Besides obvious records (eg: financial and legal records), there are many documents that may or may not be records. For the most part, it's up to each program to decide which documents are records and archive them appropriately.

Every mail is sacred

[kfg]

Every mail is great
If a mail is wasted
The gods get quite irrate

Every mail is wanted
Every mail is good
Every mail is needed
In your network neighborhood

Really, the idea of not being able to record and save every post-it note being equated with those times and places where writing itself was denigrated into virtual nonexistence is a bit silly.

KFG

Re:Why do we need to archive everything?

[felix71]

Actually, one of the main complaints Historians have is incomplete information about the past. Not having every little tidbit makes it impossible to figure out how people actually lived. History _should_ be more than just names, dates, and events. If we can properly preserve and index items that seem really mundane to us, future generations have a _much_ better chance of having some real understanding of how we developed as a society.

the more I think about it...

[1nv4d3r]

I'm not sure most of this stuff is worth making preserving digitally enough to justify the cost. Just print em out, and put them in a Raiders of the Lost Ark-style warehouse. The few people who want to see all of clinton's administration's emails can travel to it and search.

I'd much rather see those hundreds of millions of dollars invested in, for instance, making all out of print recordings and books available on-line. It's a smaller problem (sounds like), but would benefit the world much more than online copies of every government employee's timecard records.

.

strip MS HTML from Outlook mails

[rduke15]

I don't know about the NASA data sets, but they could certainly save a few petabytes by stripping the stupid HTML part of all Outlook emails...

Moore's Law saves the day

[G4from128k]

In 1987, a Mac II came with a 40 MB drive. 17 years later, a PowerMac G5 came with 160 GB drive. This was at least 4000X improvement in storage density and price (and 1987's drive was both physically larger and more expensive than 2004's drive).

Assuming we continue the current rate of advance in storage density and price, future archivist should be able to buy a 0.64 PB drive for under $500 in 2021. A mere quarter of million dollars will provide enough space for a copy of all that stuff.

I'm guessing... steady state.

[dpbsmith]

The Zapruder film was the beginning. In recent years, I've been dumbfounded by the vast extension in recording and documentation of things like crimes in progress, natural disasters, America's Funniest Home Videos, you name it. A plane crashes, and the next day there are ten different home videos from people in the vicinity who had camcorders.

I believe the cost of traditional photography in constant dollars dropped enormously between my parents' time and mine. I know we took about ten times as many silver-on-paper and Kodacolor dye-on-paper snapshots as my parent did. Then we got a camcorder. My parents captured about three hours total of 8 mm silent home movies. I have about forty hours of 8mm and digital-8 camcorder tape.

And since my wife and I got digital cameras, we've been taking five to ten times as many pictures as we did when we used film cameras.

Now, YES, I'm on the format treadmill. Got most of the old 8mm movies transferred to VHS. Got most of the VHS transferred to DVD. Got a lot of the old slides scanned. Got most of my digital images burned to CD. In the last five years, I've probably spent a hundred hours, or 0.2% of my life, on nothing but struggling to copy from old formats to new. I've spent a small fortune getting Shutterfly to print pictures, because to tell the truth I have much more faith in the prints surviving than the CD's.

So, I don't see a digital dark age. I see a bizarre situation in which the quantity of material recorded in digital form continues to increase exponentially for quite some time. _Most_ of it will get lost, and the percentage that survives, say, a hundred years will keep going DOWN exponentially with time.

But I'm guessing the total quantity of 21st century material available to historians of the 23rd century will, in absolute numbers, be just about the same as the total quantity of 20th century material.

It's one of those mind-boggling things like personal death that one can never quite come to grips with. The future is unknown, and we can accept that. But the fact that most of the past is unknown is equally true--and very hard to accept.

The Solution:

[DarkEdgeX]

NARA needs to open up tons and tons of GMail accounts. Where do I send my invites so I can contribute?

Try to help correct other's math sans sarcasm.

[jbn-o]

You were just a little over 12 times too much. Let's just hop you don't write code for a living :p [...]

To you and the countless others on /. who offer their corrections in a similar tone: Yes, we get it, the parent poster goofed and you supplied a correction. Given the trivial context here, it's hardly a big deal and doesn't warrant sarcasm. Everyone make mistakes and plenty of people make mistakes in their work every day, including people who do work where lives are at stake. That's one reason why it is good to work with other people. In life it's far more important to be forgiving, keep things in perspective, and help other people without the wiseacre commentary and then move on.

Cost-of-copy and modes of failure

[G4from128k]

I think you're missing the point, which is that all that data is now much easier to lose, especially in the short term, if it's not taken care of properly.

Perhaps, perhaps not. Sure, digital data can be lost easily, but it can also be copied/backed-up more easily. Assuming $0.01/page for paper copy (a gross underestimate of the cost of paper, toner, and labor for copies) and assuming 10 kB data/page (an overestimate), $10/GB (for high-end maintained storage), then cost ratio is at least 100:1 in favor of digital (and probably 1000:1). Inaccessible formats are a concern, but an automated batch process at the time of initial archiving can, at least, convert the data to some data format standard with a longer likely lifespan(e.g., plain ASCII, RTF, PDF, HTML, etc.)

Paper is its own single-point of failure concerns and the huge cost of copying makes those concerns real. Digital does add some new modes of failure (e.g., format obsolesce), but I think those are not as burdensome as the physical costs of copies.

entropy

[YesIAmAScript]

You can calculate the amount of entropy in a document (text or no) and that is a limit to how small you could possibly make it.

I don't recall how close modern methods like arithmatic encoding make it to that limit, but I know it's close enough that we couldn't double the compression ratio of text documents from the current state of the art.

Trellis coding is a system for dealing with induced errors in modem signalling. It allows you to cancel some of them out. It doesn't actually increase the throughput in an ideal situation.

The thing that allowed us to reach the limit for a phone line is combined amplitude-phase coding, or the creation of the "constellation diagram" for modem encoding.

The constellation defines certain combinations of phase and amplitude that represents groups of bits (a baud). Trellis coding simply defines additional combinations that are not sent. If you see any of these on the receiving end, then you realize that the constellation is either being twisted (phase error) or shrunk/grown (amplitude error) and you can try to compensate for it.

The name comes from a trellis, like you grow plants on. The legal signals sent should go through the holes in the trellis. If you receive a signal that falls on the trellis (hits the trellis) you adjust it so that it goes through the trellis and assume this adjustment factor can be used to adjust other, valid hits too to more accurately determine the data that was sent.

industrial espionage would be sillier

[mbius]

"Give them to me."

"What do you want??"

"That Gem...and the Holograms."