We've got an eLearning system that operates on two sites. On each site is
a web server distributing content. What we wanted is to have one DNS entry
(elearning.oursite.com) with one address for site_1 and another address
for site_2. This way, all elearning will come from
http://elearning.oursite.com/whatever and will always go to the local

We've put in two addresses against the record and pings from a PC on
site_1 return the address for site_1, and pings from PCs on site_2 return
the address of site_2.

Site_1 and Site_2 have their own DNS servers running on NetWare 6.5. They
service the same DNS database. The PCs on site_1 have DNS_Site_1 as their
primary and DNS_Site_2 as their secondary, and vice versa on site_2.

Now, that's really nice, but I dont know HOW it actually does it. I dont
want to rely on it unless I know its something that will always work. From
what I've read, when there are multiple addresses, it'll either return one
or the other either randomly or round robin. But in this case, it always
returns the one closest to the PC.

We set a PC on site_1 to use DNS_Site_2 and it still returned the address
of site_1, so I suspect it may be the PC (XP). However, I really dont
know, but want to.

Also, our addresses on site_1 start with (for example) 10.5.x.x and on
site_2 with 10.6.x.x. Maybe it looks at each address and determines which
one to return by octets.

Anyway, if anyone has any ideas why this is so, I'd love to hear them.