[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
I just took a look at Intermemory site, www.intermemory.org .
They have a first paper detailing the scheme, and an implementation
paper which discusses how they've built a prototype at NEC Research.
* They have a prototype implementation which breaks data up into
32 fragments, so can tolerate up to 17 failures. They're working
on another prototype which may release code outside of NEC Research.
* Their "economic model" is similar to Roger's : they think people
will donate resources to the service in exchange for the right
to store their own data. They differ from Roger in this way :
if a donor stores M megabytes, the donor receives c * M
amount of space on the system _forever_. Here c is some constant
between 0 and 1 picked by the system desingers. So once you
are in the system, you are never supposed to leave.
* Neither their first paper nor their implementation paper treat
cryptographic issues explicitly. Joe Killian is listed on their
"people" page, but he is not a coauthor for any of the papers
that I saw. They basically said "we're going to wait to look at
crypto." Much less anonymity...
* On the other hand, they spend much time considering synchronization
issues, the space requirements for sharing schemes, and other
systems issues. These are discussed in some depth, particularly
in the implementation paper.