In a nut shell...
I am in the middle of a high(well sort of) availability project my self.... i am set up as follows(so far)
4 servers in all, will expand to 6 for full redundancy..
Server 1, named Megs, has 4 IPs on it, On for web related stuff(ftp,http, https), 1 for mail related stuff(pop,imap,smtp,mx,etc) one for its self, and one spare IP.
All services point to this Megs, then redirected to 2 backend servers Lois & Peter
Lois & Peter, mount NFS(/var/www/virtual & /var/mail/virtual) to server named Joe, Joe runs Omega, and all provisioning is on there...
Lois & peter also mount /etc/apache2 in read only mode
i rsync /etc/courier files over each minute from Joe(omega machine) to peter and lois.
I use HAproxy to load balance the http to peter & Lois
each back end server has 2 300 gig mirrored drives, one partion. If one server goes done other one takes over with no interruption at all ...
I wil then add another server as a Megs 2 and Joe2, when i am finished i wil post detailed docs on how i did it....
Have a nice day..
Rob..
rbtux Wrote:i have two well equiped servers providing each mail, dns, database in isolated vm...
the webservers are always standalone... if a customer wants high availabilty i do this with two vps serving the domain and without ispcp because these setups are always customized...