On Monday 20 August 2007 21:59, Malkus Lindroos wrote:
> Matthew Toseland wrote:
> > On Monday 20 August 2007 15:48, Malkus Lindroos wrote:
> >> Just realized a problem in using average ping times to rate limit nodes.
> >> If many of your peers have their connections heavily loaded on the
> >> average, this leads to limiting your node. At the same time some of you
> >> peers may not be heavily loaded, because their connections are not
> >> overloaded. This leads to slow transfer rates in all those nodes
> >> connected to few nodes with not enough bandwidth.
> >>
> >> As a suggestion, I suggest using the minimum ping times from other nodes
> >> to evaluate the load on a given node. Trying effect of that now, though
> >> all nodes should do the same to see if it works.
> > 
> > What if they're on a LAN? On the same ISP?
> > 
> > I believe we use the median of the average pings of our peers.
> 
> Hmm, true, there is a problem if they're on the same lan and the node is
> bandwith limited and the bandwith is set too high by the user. There
> should be an option to exclude nodes on the same lan, maybe by the user?
> However, there would be no problem if the user knows his bandwith or the
> node is not bandwidth limited.

On the current code, the ping time has no relation to node-side bandwidth 
limiting. Sometimes LANs will have real internet addresses. And again, what 
if they are on the same ISP? Maybe on an eccentric configuration like a big 
LAN? The minimum ping time could very well be submillisecond if they are on 
the same LAN!
> 
> Of course it depends on what the ping time is supposed to reflect? At
> least for my nodes it seems to reflect the load of the network the node
> is connected to, not the load of the node or its internet connection.

It's a combination of the round trip time under ideal conditions (<0.2ms on my 
LAN), and the effect of the traffic on the links between the nodes (including 
rate limiting).
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: not available
URL: 
<https://emu.freenetproject.org/pipermail/tech/attachments/20070821/a08e526b/attachment.pgp>

Reply via email to