Peter Andreev wrote: > >> >> > Victor, we researched this topic and learned that response time highly > >> >> > depends on distance between user and resolver, while cache influence > >> >> > on this value is lesser. > >> >> > So I advice you to keep all as is. > >> >> > >> >> Be it so. Thank you. > >> > > >> > And the reason for the whole thread. One of the customers told me that > >> > 8.8.8.8 is faster than our own DNS servers which are located on the > >> > same 100 MBit/s LAN with them. I was shocked but it seems true, at > >> > least for the answers which are not yet cached. > >> > >> I don't know what software google uses on its resolvers, but I suppose > >> something with shared or synchronizing cache. May be they also make > >> preventive lookups on popular domains to fill this cache. And the > >> reason why 8.8.8.8 seems faster - it answered from cache while your > >> resolver made full lookup chain. > > > > Duh! That is why I started thinking about some cache synchronizing > > technique for my resolvers. > > Preventive lookups can be made via self-written scripts.
Sure, after query log analysis. > > AFAIK there is no free open source implementations providing cache > synchronization between different resolvers. Unbound cannot do that, can it? I am surprised. After all, squid siblings are quite common. -- Victor Sudakov, VAS4-RIPE, VAS47-RIPN sip:[email protected] _______________________________________________ [email protected] mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to "[email protected]"
