On Wed, Oct 22, 2003 at 12:26:06PM -0700, Mike Stump wrote: > > We have yet to establish that freenet routing does not work. > > What is the metric that indicates that it works? > > outboundAggregateRequests: 432580/day > inboundQueryRejecteds: 255612/day > resetRatio: 767 tries, 23 succcess/day > > For a total of 0.3% chance of a query getting an answer? If only 0.3% > of all queries even get a DF, certainly the chance of actually getting > all the data is even lower?
Where do you get the 0.3% chance from? psuccess is the probability of the whole requests successfully moving data. http://127.0.0.1:8888/servlet/nodestatus/psuccess_data.txt More detailed data can be obtained from requestSuccessRatio and routingSuccessRatio (I use the hourly means), in the Diagnostics page. My psuccess is closer to 5% than 0.3%... > > Lowest global time estimate: 318570ms > Highest global time estimate: 459317ms > > 459s latency, seems kinda high. Yeah... that's for full file transfer though. I know, it's bad. > > Do the odds of getting an answer have to drop below 1 in 10,000 before > we label it as a doesn't work? Does the latency have to rise about 1 > day before we so label it? Just because Freenet is not working now does not mean Freenet *routing* cannot be made to work. It has worked a lot better in the past. > > Further, if I ask fred if it works, he says the most successful node > that he talks to has a 1 in 2,792,500 chance of getting the data he > wants. Meaning, I have to burn 2.7 million queries before I get a bit > of data. How many bytes does 2.7 million queries take? > > Also, I stream out around 8KB continuous, and I get in around 100 > bytes a day on a good day, not that I try too much. It is rather > painful, as my browser only does 4 active connections at a time. It > `feels' like it doesn't work. When I click on something, I actually > would prefer if it came up. Why does your browser only do 4 conns at a time? Can't you fix that? > > Also, when it does work, it works at 1-5 bytes a second transfer rate, > that sure feels like it doesn't work. What is the theoretic cost to > being anonymous? 1000x inefficient seems slightly high to me. > > I know, to measure it, we can send out with the load information, how > many bytes to the actual user freenet gave them and how much data we > pump out the upstream connection on a global basis. We can then have > fred display E, as E=userbytes/outbytes. > > These two can be computed: > > newglobaluserbytes = myuserbytes *0.01 + globaluserbytes*0.99 > newglobalupstreambytes = myupstreambytes * 0.01 + globelupstreambytes*0.99 > > each hour. Over time, the global would tend to be the average of all > nodes near you. > > On non-anonymous P2P networks, we expect E would be around 0.99. On > freenet, we know it will be lower, but with efficient routing, and 5 > copies to obscure requester and provider, I was almost hoping for E to > be around 0.18. freenet feels like it is at 0.00001 or below. Now, I > know efficient routing may not be possible or desirable, but I was > hoping that a 10x overhead might be enough, for a total E of around > 0.018, this would be vastly better than how freenet feels, if you want > to see hard data, put in code to measure it as above, and then we can > see just how bad E really is. -- Matthew J Toseland - [EMAIL PROTECTED] Freenet Project Official Codemonkey - http://freenetproject.org/ ICTHUS - Nothing is impossible. Our Boss says so.
signature.asc
Description: Digital signature
_______________________________________________ Devl mailing list [EMAIL PROTECTED] http://dodo.freenetproject.org/cgi-bin/mailman/listinfo/devl
