MTBF

[Brad Knowles]

At 10:53 AM +0100 9/5/01, Marc.Thach at radianz.com wrote:

>  I'm running BIND 8.2.3 on Sun E220R Solaris 8.  I need to put together some
>  reliablity figures for each server.  I can get this data for the hardware
>  and OS, but what about BIND?  Has anybody out there done this before,
>  either collated real data or used some sort of estimatation technique?

	I think it would depend too much on your particular installation, 
how much RAM you have, etc....  For example, if you're trying to run 
a heavily loaded nameserver on just the internal interface (which 
typically doesn't have much buffer RAM onboard, thus requiring the 
CPU to handle a much higher rate of interrupts in order to service 
it), and you're trying to do this in just 16MB of RAM, that would be 
a totally different kind of situation from someone running a lightly 
loaded machine on an old UltraSPARC clone with an add-in high-quality 
NIC and 512MB or 1GB of RAM.

-- 
Brad Knowles, <brad.knowles at skynet.be>

H4sICIFgXzsCA2RtYS1zaWcAPVHLbsMwDDvXX0H0kkvbfxiwVw8FCmzAzqqj1F4dy7CdBfn7
Kc6wmyGRFEnvvxiWQoCvqI7RSWTcfGXQNqCUAnfIU+AT8OZ/GCNjRVlH0bKpguJkxiITZqes
MxwpSucyDJzXxQEUe/ihgXqJXUXwD9ajB6NHonLmNrUSK9nacHQnH097szO74xFXqtlbT3il
wMsBz5cnfCR5cEmci0Rj9u/jqBbPeES1I4PeFBXPUIT1XDSOuutFXylzrQvGyboWstCoQZyP
dxX4dLx0eauFe1x9puhoi0Ao1omEJo+BZ6XLVNaVpWiKekxN0VK2VMpmAy+Bk7ZV4SO+p1L/
uErNRS/qH2iFU+iNOtbcmVt9N16lfF7tLv9FXNj8AiyNcOi1AQAA