On Tue, Jul 23, 2002 at 10:52:55AM -0700, Ian Clarke wrote: > On Tue, Jul 23, 2002 at 05:49:26PM +0100, Matthew Toseland wrote: > > Hmmmm. If we're going to do selective caching (which we haven't yet > > done, and which is pointless until we fix the datastore bugs), we > > probably should involve the success probability of the key in the > > calculation; this is the number of successful requests over the total > > number of inbound requests in a given keyspace segment (there are 256 > > segments), and is calculated for overload triage now... > > Again, I am not sure I like this - we have seen no evidence to suggest > that the natural specialization mechanism requires such forceful and > inelegant assistence to ensure efficient operation of the network. I > would strongly disagree with implementation of this until it has been > simulated. Okay, what gets put in the datastore will naturally be biased according to specialization, so maybe we don't need to do this... size bias will probably be necessary though. > > ian. > > -- > Ian Clarke [EMAIL PROTECTED] > Founder & Coordinator, The Freenet Project http://freenetproject.org/ > Chief Technology Officer, Uprizer Inc. http://www.uprizer.com/ > Personal Homepage http://locut.us/
msg03479/pgp00000.pgp
Description: PGP signature
