Okay, for anybody that missed my 2c worth yesterday (or day
before...I can't even remember what day is what at the moment), I am thinking
about writing a HOWTO but only if:
1. There is enough support for it; and
2. There is no good reason for me not to bother.
OK, enough of the blabbering. I have just finished
implementing a new system that allows e-mail around an entire city (further with
no major problems except cost of interstate and international communications)
even though all e-mail only goes to one domain name. The e-mail is distributed
via NFS to shares that connect when the remote location (either permanent ISDN
or dialup modem) dial up, and there is no problem with messages not getting
delivered when they are not connected, they are simply deferred until they
connect.
Users can check their e-mail from virtually anywhere in the
world, no matter if they are inside their work network, or outside on the
Internet, via the ability of distributed NIS. The idea of this system is to
dramatically decrease network traffic by not having unnecessary bounces, having
POP servers onsite, and the network will still appear to work (from any
location), even if the main server goes down. Cost is also decreased as all that
is required is a small server (x86's do the job fine) in each location, and one
main server that only really needs to handle traffic and requests, with no real
data storage capabilities. Messages are simply deferred if a server is down, but
POP mail can still be retrieved. If one link goes down, nobody else is affected
in the slightest.
If anybody wants a HOWTO on this (ie you have a use for it),
then please e-mail me personally, or post your comments on why or why not this
is useful to the list. If you have already contacted me then please don't bother
again!
Thanks for your time
Brett Randall.
|