On Fri, Jan 18, 2002 at 02:36:28PM -0800, Alvin Oga wrote: > > hi ya > > you can try to do a stress test on your server from the outside world > and see if its response time is still acceptable > > or simple way... > - when yur ids/monitoring tools says that the server doesnt > exist or doesnt respond... you know its too heavily loaded > and it'd be time to make "www" into a cluster of loadsharing > servers > > > Hello, > > I have a Dell PIII 800 MHz with 256 Meg Ram running debian and apache, > > along with mysql and sendmail. I have a redundant t1 connection to this > > server. > > I have a website that gets about 800000 hits per month and transfers > > about 8403274 KB per month. > > The site gets an average of about 922 Hits per hour (average of 22139 > > hits per day) the daily transfer rate is an average of 233,424 KB per > > day.
Add more RAM!! :) With that thing you should have at least 512MB. With today's cost of RAM - it's pennies.. I would recommend more than 1GB if you expect your traffic to increase a lot. Anyway, on a T1 connection, I do not think that it is capable of killing a machine like that unless you use a lot of SQL things and non-static things or many mailing lists.... In all cases more RAM is good.. I would suggest a cluster only if you want redundancy - for raw performance just get a mobo with 2, 4, or 8 processors... But generally, disk access is _bad_ for overall performance of the server. So having as much RAM as possible is a must for optimal caching. - Adam PS. Back in the olden days, a 486 66MHz could handle 1 hit/s for static website.. That's >80k hits a day..
