On Tuesday 27 Mar 2012 19:31:13 Robert Hailey wrote: > > On 2012/03/27 (Mar), at 12:06 AM, Steve Dougherty wrote: > > > I hope to investigate that this summer as part of Google Summer of > > Code! Are there metrics to focus on? > > The most prominent would be: > * CHK success rate from seed nodes should be 100% (or at least no different > from a common non-seed node) > * there should be no "classes" of seed nodes (e.g. busy ones working & > lesser-ones not) > > There are also less clear indicators: > * announcements prematurely ending through some seed nodes > * announcement rejection rates > * network fracturing/reachability > > > Would you be willing to contribute analysis, > > I'd be happy to detail the theory and supply a patch (though both might be > more readily available through the list archives)... > > In a nutshell, it amounts to the "valid" paths (from the seed nodes into the > network) by virtue of being traveled by the announcement are consumed by the > invalid/fledgling nodes (as that's where they get hooked in). The depth-first > notion intends to solve this by replacing the "deep" connections and not the > shallow ones that the seed nodes rely upon (and has a strong theoretical > analog to insertion requests).
If it can be tested empirically we should try this (although probably we want more data gathering first). -------------- next part -------------- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 198 bytes Desc: This is a digitally signed message part. URL: <https://emu.freenetproject.org/pipermail/devl/attachments/20120330/f44eee34/attachment.pgp>
