On Tue, Nov 28, 2006 at 03:50:04PM +0000, Michael Rogers wrote: > Jano wrote: > >Aren't hits prioritized somehow? It seems of paramount importance not to > >drop a hit. > > True, but the data can be relatively large (32 kbytes) so they also > constitute the majority of traffic. It's important not to let them time > out, but they need to be interleaved with other traffic.
They probably don't constitute the majority of traffic. Forwarded requests use just as much bandwidth, on the rare occasion that they are successful, and successful forwarded requests are almost by definition more common than local hits. > > >I know this is very basic, but bear with me: Do hits follow the inverse > >path > >than their respective queries, or is there something else going on? > > They follow the reverse path. (I'm not sure whether this has been > implemented yet, but there was also a proposal to trigger reinsertions > of the data at random points along the path, in which case the data > could spread off the path.) That has been implemented. Simulations would be welcome eventually. > > Cheers, > Michael -------------- next part -------------- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 189 bytes Desc: Digital signature URL: <https://emu.freenetproject.org/pipermail/tech/attachments/20061128/0c1622f6/attachment.pgp>
