high availiblity DNS

[Barry Margolin]

In article <4.3.1.2.20000625220231.01cf3c00 at pop.enteract.com>,
Ramin K  <ramin at badapple.net> wrote:
>Availability, I think we can handle internally, what's bothering me is the 
>sheer volume of requests we need to keep up with. Sorry I didn't really 
>make that apparent in the email. We are already seeing drops and we've only 
>moved 15% of the traffic over to our own DNS system. Rather then buy 12 
>more machines we're looking at using the existing servers to the limit of 
>their CPU/RAM capacity.

The root servers probably get many times the rate of queries you get, and
they get by with a single named process, AFAIK.  If you're seeing drops, it
may be due to the kernel running out of UDP buffers, and increasing the
number of named processes isn't going to help this.  All you're going to
accomplish with multiple processes is that they'll be fighting against each
other for shared system resources, slowing each other down rather than
making effective use of all the CPUs of the system.

-- 
Barry Margolin, barmar at genuity.net
Genuity, Burlington, 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.