Jan Gyselinck wrote: > No you can't. I know Bert was toying around with a couple of ideas > in that direction, the main problem is the fact that checking locally > (before fetching it from the right location) costs time and thus > increases the query-response latency. In practice the problem seems
Well if you could do a parallel query to all siblings (similar to squid), and after say 20-25ms of getting no replies you go direct, chances are over time you could find that with the law of averages the amount of time saved verse lost with this method more then adequately makes up for any initial losses. Not to mention when I do DNS traces with dig, won't always be a true reflection I know, it's not uncommon to see times over 200ms as not all gTLD servers are on anycast IP space, which is stupid but another argument altogether. > to be a moot point, you surely are not the only one with a load > balancer setup with x amount of servers behind it, and the other open/free > big DNS resolver implementations don't support 'peering' either. Is there anything inherently bad about doing something similar to squid's notion of siblings/parents in DNS? It seems to work well as far as I've seen. -- Best regards, Duane http://www.freeauth.org - Enterprise Two Factor Authentication http://www.nodedb.com - Think globally, network locally http://www.sydneywireless.com - Telecommunications Freedom http://e164.org - Because e164.arpa is a tax on VoIP "In the long run the pessimist may be proved right, but the optimist has a better time on the trip." _______________________________________________ Pdns-users mailing list [email protected] http://mailman.powerdns.com/mailman/listinfo/pdns-users
