Hacker News Reader

Comment by bombcar | Hacker News Reader

I think people don't really stop to consider - the Internet Archive could be gone tomorrow; either through legal action, malice, accident, mismanagement of the non-profit, etc. It's just considered "an Internet tool that exists".

I would argue that their attempt to even try this emergency library thing indicates that the people in charge are a bit too cavalier with what they have, for whatever (likely well-intentioned) reasons they may have.

And while individuals can mirror/datahoard the publicly facing parts of the IA, that's not all they have access too.

View on HN

So, we need a distributed planet-size, multi-redundant, partly-encrypted, archive that can absorb and host the entirety of IA multiple times (for redundancy), INCLUDING the borrowing library.

Yes. If you have the pockets and logistics behind you, I'm sure Brewster would invite you for lunch to discuss. You'll need people for infra ops, people to coordinate hosting racks across the globe (preferably on every reasonable continent), and whatever the current cost is of a few exabytes of cost efficient storage hardware.

This comment reminded me of famous quote: "When someone says "I want a programming language in which I need only say what I wish done," give him a lollipop."

I didn't say it'd be easy. In fact, I was being sarcastic.

Sadly, my doctor also said I can't have the lollipop.

"why build one when you can have two at twice the price"

they have a duplicate headquarters in Vancouver now, a similarly grand building to their SF headquarters. Years ago there was a collab with the library of Alexandria in Egypt to host an offsite backup but I don't think it panned out.

This would be the poster child for durable backups such as Microsoft Silica (if Microsoft dared to piss off those plaintiffs).

At 99 Petabytes, an offline copy would take about 2,000 LTO-9 tapes. I'm not familiar with other vendors, but a single IBM TS4500 tape library offers about 400 PB of near-line storage and I don't think IBM would be making the largest ones in existence.

Also, CERN could host multiple copies on unused blocks of their storage farm.

edit: just found a StorageTek (now Oracle) that can do "57.6 EB of uncompressed data". That's just surreal. HPE sells a much more modest unit that can store 2.5 EB.

Reply on news.ycombinator.com