> Some applications (Frost) need to generate more that one guessed key every 
> second. That means in order to accommodate these applications, we MUST 
> implement TUKs.
We must talk about these new TUKs you go on an on about someday.  Could you give it to 
me in a
nutshell?  It seems like what Frost wants is board type object, something were 
serveral values can
be written to the same object, and when you request, you get all of them back.  I read 
a paper
proposing something like this.  I'm not sure how relevant it is.  Shall I dig it up?

> > I suppose an adversary with a certain fraction of nodes would probably
> > still be able to use the keys already mapped to a target area to do
> > dammage.  An advesary with f nodes compromised would get about f/N
> > *<arverage route hops> of the messages in the system(for f<<N).  If he
> > collected all the keys in the target keyspace and requested them all
> > repeatedly from all over the place, it might overload the area of the
> > network responsible.  Caching would help some.
> 
> The data is actually there, so it would not make the nodes look bad. Caching 
> would help a LOT.
Sure they wouldn't look too bad, but they'd still might be overworked.  Imagin if 
someone
requested everything from your 40GB store.  It could take days before you have it 
uploaded to all
your neighbors.

> > There's still the SSK attack, but you can fight this with a some caching
> > and a realistic limit on the update rate.
> 
> What attack is this?

The SSK attack:
* pick an SSK close enough to the target hashspace
* insert different versions of 32K junk to it.  
You know all that bandwidth is going to the same place.

I think this can be solved.  Just limit the number of updates you can make under an 
SSK to say 1
per hour and check the dates on the insertion messages.
If the datetime on the message is in the future drop it.
If its less than date on the version I have cached, keep it but delay the insert for 
an hour. 
This should let caching help defend.


__________________________________________________________________

Gesendet von Yahoo! Mail - http://mail.yahoo.de
Logos und Klingelt�ne f�rs Handy bei http://sms.yahoo.de
_______________________________________________
Devl mailing list
[EMAIL PROTECTED]
http://dodo.freenetproject.org/cgi-bin/mailman/listinfo/devl

Reply via email to