Using a low TTL to enable a fail-over cluster?

[Barry Margolin]

In article <a7g263$jq0 at pub3.rc.vix.com>, Nate Campi  <nate at campin.net> wrote:
>
>On Fri, Mar 22, 2002 at 07:20:41PM +0000, Barry Margolin wrote:
>> In article <a7fs1h$irb at pub3.rc.vix.com>, Nate Campi  <nate at campin.net> wrote:
>> >
>> >In reality, BIND and it's usage of RTT to find and use the fastest
>> >responding nameserver makes this incorrect. Since most of the
>> >nameservers out there are BIND, this has a visible effect on the
>> >distribution of queries to most nameservers.
>> 
>> Only if the response times are significantly different.  It uses fuzzy
>> comparisons of the RTT's, so if all the servers respond within a few
>> percent of each other, they'll be treated equally.
>
>Hence the usage of the word "most", as in "this has a visible effect on
>the distribution of queries to most nameservers".

I thought you were using that to distinguish BIND vs. non-BIND servers.
But if the authoritative servers all have similar response times, the use
of RTT won't have a visible effect on BIND servers.

-- 
Barry Margolin, barmar at genuity.net
Genuity, Woburn, MA
*** DON'T SEND TECHNICAL QUESTIONS DIRECTLY TO ME, post them to newsgroups.
Please DON'T copy followups to me -- I'll assume it wasn't posted to the group.