On Fri, Apr 20, 2001 at 03:44:50AM +1200, David McNab wrote:
> > Do we even have 100 nodes on the network?
>
> Now I'm _really_ getting worried.
Well, I think that Fred is working, so clearly we do (it's possible to
have 100 htl go by on a smaller network because of backtracking, but
still). The fact that you had to wait many minutes for a response would
indicate that it really did make all those jumps.
> As I was tossing and turning in bed earlier tonight, contemplating this
> problem, I thought that the issue might be the opposite - thousands of
> nodes.
>
> I realise I'm still painfully ignorant of Freenet's topology
> But I wondered if Freenet is getting so vast that a HTL=100 insert is just a
> mere drop in the ocean.
Well, obviously the idea is that should not be happening. Even on if the
routing were completely random, the chance that two selections of 100 (the
effect of a randomly routed insert and a randomly routed request) share a
node would only fall to 30% around 30000 nodes or so - which I severly
doubt there are. So clearly, the problem is not that the network isn't
scaling - it's completely broken.
> I clearly remember Ian saying that "a new node registers with inform.php on
> average every 15 minutes". That's 672 nodes per week. Surely everyone can't
> be rebooting all the time! Especially since a good percentage of users would
> be linux based, lacking the need to reboot every 5 minutes.
>
> My mind churned with ideas like 'multiple insertion points' - say, doing
> HTL=5 inserts to each of (say) 50 to 250 nodes picked randomly from
> inform.php, and whether that would be worthy of consideration as standard
> insert procedure.
As I told GJ, if we don't believe that we can make the algorithm work we
are better off abandoning it then trying to pretend it does by trying to
force things to be visible.
<>
> But, is there really such a low number of nodes? 100 or less??!?
> Surely not!?! But if so, then there's *real* cause for worry!
> If that's the case, then the Freenet architecture as it stands might have
> real problems.
The Freenet architecture as it stands has real problems. This is not news.
> I'm in the dark here.
> Could someone please clarify.
>
> Cheers
> David
--
'DeCSS would be fine. Where is it?'
'Here,' Montag touched his head.
'Ah,' Granger smiled and nodded.
Oskar Sandberg
[EMAIL PROTECTED]
_______________________________________________
Devl mailing list
[EMAIL PROTECTED]
http://lists.freenetproject.org/mailman/listinfo/devl
