Re: [Help-gnunet] INDIRECTION_TABLE_SIZE and download speed

[Igor Wronsky]

On Fri, 6 Sep 2002, Tracy R Reed wrote:

> That would not explain how the transfer starts off so fast. The consistant
> patterns seem to be that either the transfer never starts (which is odd
> because the file showed up in search results seconds earlier so it must be
> out there somewhere right?)

Nope. It doesn't have to be anywhere. It suffices that someone
has made the search previously when the file was there, making 
the search result get off the indexing node. Then the indexing 
node might go away, but the "header" stays in the network 
because its just like any other block. :( 

> pulled over a slow modem, who knows. It just seems odd that I am yet to be
> able to retrieve any more than one or two meg of a file, usually only a
> few tens of k.

The current exponential backoff routine without ceiling might 
be overkill, but on the other hand we wouldn't want to flood
the network with queries for something that is not there. Some
compromise would probably be in order.


I.