----- Anonymous ----- 2007.04.29 - 03:05:11GMT ----- You are only halfway there. Someone needs to download it now. As I recall, the last race results for the 0.5 network were 18 hours from the beginning of the insert to the completion of the download.
There are some things to think about while the second part of the race is taking place, you are the developer of this network. If there was anyone who would know how to tune their node for best performance, it would be you. Another thing is that you most likely have many long established connections to others who also have many long established connections. Think about the performance of the newbie -> newbie -> newbie -> newbie nodes. I see the 0.7 network to be like a pyramid scheme. The ones who got here first have the best connection because they have been able to weed out the low performing nodes, while the next level gets stuck with the low performing nodes, and it gets worse from there until the data from the newest user has to go through several levels of poorly performing nodes to finally reach you and your clique of fast nodes. ----- Zeren at Ev72+T5cnPcLHRRiLYEjAt+8YQ8 ----- 2007.04.29 - 15:05:24GMT ----- Isn't that what you would expect from a network seeking security via detachable clusters of mutually-trusted nodes? Security will be enhanced for those more established. New nodes initially need to behave well during an evaluation period connecting largely to other nodes of unknown trustworthiness. Since those more established will tend to be more long-term, their connections will be stronger and the need to risk new connections will be less. Yes, this does mean new nodes will initially see mostly other new nodes with a high turnover. It also means that when a massive assault on the network occurs, detached proven clusters or groups of clusters could survive in relative isolation for the time necessary. That said, during relatively safe periods, we should also to try to connect to new nodes as much as possible to help new nodes become established. ----- toad at zceUWxlSaHLmvEMnbr4RHnVfehA ----- 2007.04.30 - 12:58:54GMT ----- What this means is that new nodes are slow, old nodes are fast. Aristocracy, like the first poster said. We talked about this on 0.5 too. Dealing with newbies (who generally are new nodes, with high turnover i.e. high likelihood of leaving) well is a problem. Opennet doesn't *necessaily* solve the problem; on 0.5 it took a long time to get good connections. One advantage is that nodes with true darknet connections are fast. Implementing opennet would: - Force old nodes to connect to newbies. But not necessarily level the playing field. - Increase the proportion of newbies who decide to stay. - Reduce the need to get new nodes manually. - Reduce the performance incentive to get new darknet nodes. - Grow the network fast. -------------- 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/devl/attachments/20070430/6f241912/attachment.pgp>
