On 2010/10/29 (Oct), at 6:22 AM, GitHub wrote:

> From: toad
>
> This will not be merged, period.

It needs a little more work anyway :)

> We do not make significant changes to routing without detailed  
> simulations. The Least Recently Used policy used for opennet has  
> been extensively simulated and while it is not proven, there is also  
> a very strong mathematical basis for it. There is every reason to  
> think that it should perform well, in other words, and automatically  
> establish a small world topology. Plus, it trades off performance  
> against location in a way which is simple and avoids any need for  
> extra layers of performance evaluation separate to optimising  
> locations.

You might be right if LRU was the only factor. However, I think that  
the announcement algorithm accounts for 95% of peer selection.

In my experience, nodes announce... get peers at the given location...  
and then are forevermore content with the announce-gathered peers. LRU  
would only have the effect that you state if we routinely dropped the  
lowest peer (in such a way that they could not just reconnect).

> Finally, I don't believe routing is the problem limiting performance  
> on the current network. The distribution of incoming requests is  
> usually very specialised, for example.

Incoming requests are specialized, that's true! But this indicates  
that *those requests that get to us* are specialized. The overall CHK  
success rate is a better measure of network health IMO.

It's too bad that there is not a way to experiment on the whole  
network without negatively effecting it; e.g. it would be *very* bad  
if a routing change prevented update-over-freenet. I guess we could  
run two parallell networks, but would require reduplicating much code.

--
Robert Hailey


Reply via email to