Average (over a month) time in milliseconds to fetch a 1MB no MIME type file from a freshly bootstrapped node (which was inserted by another freshly bootstrapped node), by year and month:
2008 08 117836 2008 09 114260 2008 10 103211 2008 11 140962 2008 12 134591 2009 01 100626 2009 02 157500 2009 03 99357 2009 04 91440 2009 05 71206 2009 06 79331 2009 07 79612 2009 08 96493 The trouble is I'm having great difficulty connecting this to code changes. FOAF went in in September. There are a bunch of important bug fixes in December. We experimented with reducing request latency, and implemented turtling, in January. Nothing at all happened in March. DB4O was merged in April. 0.7.5 was released in June. Caching changes in end of July, but it's a limited sample anyway, give it time to settle imho. PROBEALL: actually seems to be working and producing consistent results without steep jumps. It shows 2229 nodes when I ran it recently. This is an estimate for the number of online nodes at a given time. Last time I ran it much, maybe a year ago, it consistently showed around 1000 - although for a while it had big jumps i.e. was grossly inaccurate (presumably due to backoff problems). If we assume the figure of 1000 is reasonably accurate then the network has doubled in size and performance has improved considerably. However, the other side of this - we have 22K downloads over the last month, that's 700 per day! And the number of nodes online at a time is only 2200. So we still have massive churn. :( Also the uninstall survey has dried up, there are very few responses now, too few to be useful. Maybe we should always show it if opennet was enabled? Or would even that be too much? -------------- next part -------------- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 835 bytes Desc: This is a digitally signed message part. URL: <https://emu.freenetproject.org/pipermail/devl/attachments/20090807/8b15ecb0/attachment.pgp>
