BIND 9.3.1 Cache Server Optimization

[Thomas Schulz]

In article <d6akho$51a$1 at sf1.isc.org>, mayer <mayer at gis.net> wrote:
>----- Original Message Follows -----
>> Hi,
>> 
>> I'm setting up a DNS cache server which will serve all our
>> customers.=20
>> 
>> The server is built with BIND9.3.1 on Solaris. Searching archivement
>> of this list, I disbale IPv6 in source code compilation and start
>> named with -n option; with BIND configuration, cleaning-interval is
>> set to 0.
>> 
>> To monitor the performance & successful rate of this server, a program
>> 'dig' two domain names ( www.nba.com, www.the-sun.com )20 time per 5
>> minutes. It is found that the successful rate of the new cache server
>> is much lower than some other Cache servers. In past 48 hours,  the
>> new cache server failed to answer 5.93% resolution requests of
>> www.nba.com, while the figure of some other cache server is only
>> 1.46%.
>> 
>> Would anybody do some recommendation to improve the successful rate?
>> 
>Nothing. Those DNS records have extremely low TTL's so it constantly has
>to go back to the authorative server to requery it for its A records.

Were the tests run simultaneously?  Your test relies a great deal on the
performance of the authoritive servers for www.nba.com.   If the tests
were run at different times, you could be seeing variations in the
performance of the authoritive servers over time.  There could also be
differences in packet loss over the internet at different times.


>Danny
>> Each word will be highly appreciated.
>> 
>> Joe
>> 

-- 
Tom Schulz
schulz at adi.com