Ian Clarke <ian at freenetproject.org> writes:

> I see that a number of people are keen to find a way to better-use nodes 
> which may not be up 24/7, but are up a lot.
> 
> One solution is for each node to maintain some metadata with their 
> reference, basically 24 values, each the probability that they will be 
> online in each hour of the day (relative to GMT).
> 
icky.  Quite icky.

> This data could be passed around with the nodes reference, and factored 
> into the contact probability for that node.
> 
Accepting data like that from other users is a big danger.

> I assume that many nodes that are not up 24/7, will be up at reasonably 
> predictable hours of the day.
> 
If you want to make this assumption, it'd be much better for the node
to just maintain data about what times the node is contactable and
factor that into routing.  But really this isn't what you want; if a
node is up, you want to keep using it since it's up; you don't want to
skip sending it 50% of requests because it's usually down at that
time.

> Having said that, I am really not sure if this is worth the effort.
> 
> Ian.

The way I'd do this would be to just move nodes' references offline
instead of removing it from the store for being down.  And every hour,
check to see if any of the offline nodes are back up, and if they are,
just put their refs back in the routing table.

Thelema
-- 
E-mail: thelema314 at bigfoot.com                        Raabu and Piisu
GPG 1024D/36352AAB fpr:756D F615 B4F3 BFFC 02C7  84B7 D8D7 6ECE 3635 2AAB

_______________________________________________
devl mailing list
devl at freenetproject.org
http://hawk.freenetproject.org/cgi-bin/mailman/listinfo/devl

Reply via email to