Skip to content

Instantly share code, notes, and snippets.

@public
Last active August 29, 2015 13:57
Show Gist options
  • Save public/9597507 to your computer and use it in GitHub Desktop.
Save public/9597507 to your computer and use it in GitHub Desktop.
Why opportunistic encryption is useless

Cisco estimate that in 2017 the internet will be doing about 1.4 zettabytes/year. This is around 355 Tb/s. Obviously it'll be spiky though, so let's round that up to 1000 Tb/s.

A Nvidia GTX 295 can decrypt AES-128 at about 50 Gb/s and originally retailed for about $500 giving you about 100 Mb/s per 2009 $1. It has a TDP of about 300W, lets call it 500W for a full system. You have the rest of the computer to purchase too so let's call it 50 Mb/s per $1 after assembly of a $1000 node.

1,000 Tbps / 50 Mbps = 20,000 GTX 295 in $1000 computers.

20,000 * $1000 = $20,000,000 USD

20,000 * 500 W = 10 MW 10 MW * 0.17 USD / KW/h = ~$15,000,00 USD/year

Add some data centres and associated junk and in 2009 you could decrypt the entire 2017 internet at line speed for less than $100 million USD. Without even thinking about using custom designed ASIC hardware or anything else you'd expect from a defence project of this scale.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment