Skip to Content.
Sympa Menu

bluesky - Re: The Grapevine Project

bluesky AT lists.ibiblio.org

Subject: Global-Scale Distributed Storage Systems

List archive

Chronological Thread  
  • From: Anthony Jones <ajones AT clear.net.nz>
  • To: "Global-Scale Distributed Storage Systems" <bluesky AT franklin.oit.unc.edu>, Wes Felter <wesley AT felter.org>, grapevine-devel AT lists.sourceforge.net
  • Subject: Re: The Grapevine Project
  • Date: Fri, 7 Dec 2001 12:16:15 +0800


-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On Friday 07 December 2001 11:29, Wes Felter wrote:
> I'm less concerned about what happens to me when my Grapevine node runs out
> of disk space than what happens to the rest of the network. Is it an attack
> if I intentionally starve my Grapevine node for disk space? I'm guessing
> the answer is no, but I have to ask.

Wes,

That is a very good question. Hopefully this will be a good answer:

The big thing with P2P is that you could sabotage your own node. It's just
that the chances of it making a difference to the network as a whole is very
small.

1.
It is relatively well protected against trying to insert large amounts of
data in a certain area of keyspace.

Large data is stored in the form of a Freenet style CHK that is you hash the
contents of your data and store it using that key. Redirects from a human
readible text string are keyed from the hash of the text string.

This means that you can't just pick an area of keyspace where you want to
insert the data. Even if you can insert one file in a particular area of
keyspace you would need to insert a large number of files in a small area of
keyspace to run it out of disk space.

2.
A Grapevine node needs to store all the files in it's keyspace (or at least
know whether or not they exist so it can go looking for them). It would be
best if they don't have to go looking for the files. It would be very bad
for performance. As I was saying before - in addition to the permanent store
(which may need some kind of expiry) there will be a cache. The size of the
cache could be reduced should core disk space requirements increase.

I just don't know what the disk space requirements are going to be. Nobody
does. The thing is that the more nodes, the less disk space each node needs
to have. Unique data is only replicated a certain number of times - this
doesn't increase with network size. Of course popular will get cached on
more nodes around where the data is centred.

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.6 (GNU/Linux)
Comment: For info see http://www.gnupg.org

iD8DBQE8EEKVhwVaoilFPn0RAmrhAKCNfszoaZ5f51Y+WbtJn4PJGhfpQgCgrI1v
+lw2PyY3GvP47LE2+1mSTW4=
=Lxnj
-----END PGP SIGNATURE-----




Archive powered by MHonArc 2.6.24.

Top of Page