Brandon Gillespie <brandon@tombstone.sunrem.com> writes:
% Because 'www' is not a part of the protocol, it is the name of the
% machine. 'www' could as easilly be 'fred.whatever.com', where 'fred' is
% the name of the machine within 'whatever.com' domain. However, it became
% an unrecognized standard to name web machines 'www'. Somebody could just
% as easilly make the machine answer to 'whatever.com'.
There is a technical reason why we have http://WWW.system.org.
It is called "service abstraction".
When you want to move services from machine to machine, you don't
want to have a change in name each time. In addition, if you are
an active site, you want to distributed the load of your services between
many machines, rather than on one machine.
We have locally adopted the convention:
ftp.site.org
mail.site.org
www.site.org
maillist.site.org
gopher.site.org
This way, we can have the services on any arbitrary machine. For instance,
we may have gopher.site.org and www.site.org be the same machine,
but because our ftp traffic is high, need to split the ftp site into
another system.
As you notice above, our mail and our mailing lists can be on different
machines. We have some LARGE mailing lists that cause trouble when
some remote host dies. We don't want local mail effected by that. So
we split the services.
If we did not abstract the name, then we would have to change the name
everytime we moved the service between machines. With the name abstraction,
you don't know (and should not care) which physical machine the service
is on. That way when a new service is being build, we can use
new-www.site.org, and when it is ready and tested, make a change in the
name server, and the change is transperant to the world....
-----
John Sechrest . Helping people use
Executive Director . computers and Internet
Computer Science Outreach . more effectively
303 Dearborn Hall .
Oregon State University . Internet: sechrest@cs.orst.edu
Corvallis Oregon 97331 . (503) 737-5562
. http://www.csos.orst.edu/