In your previous mail you wrote: > > Keeping state for bazillions of DNS TCP connections to a resolving > > server will present further challenges. > > Only the DNS people think that. The HTTP people are used to many TCP > connections to manage and do not think it is impossible.
=> I could agree (I've just developed a perf tool for DNS over TCP so I can confirm all the burden/limit is for the kernel) but in the DNS the server must not close the connection just after the response so bad clients (i.e., clients which never close connections and leave them to timeout in middleboxes) should stress more a DNS server tahn a HTTP one. Anyway the initial idea doesn't make sense: it is easier to shutdown an open resolver than to make it to use TCP. Regards [email protected] PS: max perf is ~20k qr/s but it is very OS/hardware dependent and in fact is more TCP connection than query/response per seconds. _______________________________________________ dns-operations mailing list [email protected] https://lists.dns-oarc.net/mailman/listinfo/dns-operations dns-jobs mailing list https://lists.dns-oarc.net/mailman/listinfo/dns-jobs
