http://www.coverfire.com/archives/2008/03/25/a-new-way-to-look-at-networking/ Thoughts and musings Sun, 15 Jan 2012 21:25:04 +0000 hourly 1 http://wordpress.org/?v=3.3.1 http://www.coverfire.com/archives/2008/03/25/a-new-way-to-look-at-networking/comment-page-1/#comment-190014 Mike Burrell Wed, 26 Mar 2008 14:23:36 +0000 http://www.coverfire.com/archives/2008/03/25/a-new-way-to-look-at-networking/#comment-190014 If you've ever looked at <a href="http://freenetproject.org/" rel="nofollow">Freenet</a>, the idea of identifying data there is very similar. Freenet aims for total anonymity, so you can't tag people or nodes, only data. Freenet is sort of a global namespace with the basic datum being the CHK (content hash key) where the "filename" (URL) is a hash of the data's contents. You gain local namespaces through a PKI. I bring this up because I always thought it was a brilliant model for dealing with redundancy in, for example, the web. Any time any two people are requesting the same data, they're requesting the same file, even if they don't know it. This has great implications for caching. If web pages were more modular (e.g., a web page wasn't just one monolithic document, but composed of smaller documents) this would be even better: all of your static content could indeed be static. Of course the problem with this sort of filesystem is the birthday paradox. I prefer to live in my dream world where hash collisions never happen ;) If you’ve ever looked at Freenet, the idea of identifying data there is very similar. Freenet aims for total anonymity, so you can’t tag people or nodes, only data. Freenet is sort of a global namespace with the basic datum being the CHK (content hash key) where the “filename” (URL) is a hash of the data’s contents. You gain local namespaces through a PKI.
I bring this up because I always thought it was a brilliant model for dealing with redundancy in, for example, the web. Any time any two people are requesting the same data, they’re requesting the same file, even if they don’t know it. This has great implications for caching. If web pages were more modular (e.g., a web page wasn’t just one monolithic document, but composed of smaller documents) this would be even better: all of your static content could indeed be static.
Of course the problem with this sort of filesystem is the birthday paradox. I prefer to live in my dream world where hash collisions never happen ;)

]]>