* Malkus Lindroos <malkus at iki.fi> [2007-08-21 17:56:03]:

> Florent Daigni?re wrote:
> 
> > Using the median insteed of the mean solves that problem... Moreover
> > really high ping times are not possible because there is a timeout.
> 
> It just doesn't seem to work like that in practice? The 0.7 node I have
> is only doing a fraction of the work it used to in 0.5 measured by
> transfer rates, cpu usage, disk usage

Consuming resources isn't a proof of efficiency.

> which also seems to reflect
> directly to the speed of the network experienced by the user.

Any precise metric ?

> 
> >> Because of this, there should at least be an option to disable the use
> >> of maxpingtimes in the config for now and see if the network would
> >> become faster.
> >>
> > 
> > Again, it's not a problem unless most of your peers are missbehaving...
> > And as you choose them it's up to you.
> 
> Well, call it misbehaving if you like, but the problem seems to be that
> my peers have their ADSL lines congested and they have high ping times
> because of that.

Mines don't

> But I don't see why this should limit my node -

Because transferring to them data they can't handle is useless ?

> especially because there are _some_ peers that are not heavily loaded,
> which my node could be using for routing while other are overloaded.

What you're suggesting here is called missrouting.

> I have tried with some success cutting of nodes with too high average
> ping times - i.e. nodes for which their owner has set a too high bw
> limit or which are too loaded.

Well, you're supposed to connect to your friends ... Educate them so
that their line isn't overloaded anymore and your ping times are lower.

> Unfortunately that doesn't work for
> opennet - and it doesn't seem good for the routing purposes of darknet
> either. 
> My friends just don't understand about DSL line speeds etc.

I won't comment regarding opennet.

NextGen$
-------------- 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/20070821/379941da/attachment.pgp>

Reply via email to