The CSO nameserver provides an efficient database for the storage and
retrieval of a variety of information over the Internet.  Its primary use
is for telephone and email directories, but it may be used to store any
type of information. 

CSO is the informal name given to an electronic phonebook/nameserver
database developed at the Computing and Communications Services Office at
the University of Illinois by Steve Dorner and others, and since adopted
by a number of other institutions. The database follows the client server
model; the server maintains the actual data and runs a program called qi
(query interpreter) that receives requests and sends back information. The
client runs a program (often called ph) that sends requests to the server.
The ph client has been ported to most major platform in use on the
Internet, from Unix to Mac and PC.  Client functions are also built into
many of the programs used to provide friendly interfaces to the Internet,
such as gopher, World-Wide Web, and their associated clients (lynx,
mosaic, etc.). 

The database is loosely structured and keyed only on people's names and on
their alias, which is a unique identifier for their entry, or on other
"Indexed" and "Public" fields specified in the server configuration; this
permits fast lookup by name or alias, but not lookup by other criteria,
such as phone number.  The server can limit the number of hits returned,
to make it difficult for people to use it to make handy mailing lists.
Most fields of an entry can also be concealed from public view. 

The database can be used to store assorted information in addition to
people's ordinary phone book information, and can be used to store
non-people information as well, such as weather data. An email address
might be registered in the database for a person; in this case the
nameserver performs the additional service of routing addresses of the
form fullname@uiuc.edu to a registered 'real' address such as
username@ux1.cso.uiuc.edu. It permits one to have a single email address
at an institution, regardless of the number of host accounts they have.
This conversion is actually done by yet another client called phquery and
not the nameserver itself.  The nameserver in effect stores the mapping
between (alias || callsign || name) and (physical email address).  phquery
is designed to be called by sendmail, perform the lookup, then re-invoke
sendmail with the new address obtained via nameserver lookup. 

A CSO database can also be used for security purposes, to validate entry
to important services such as dial-in terminal servers. The present
implementation of the terminal server security software (qtacacsd for
Cisco terminal servers) does indeed query the nameserver. 

At the University of Illinois, UofI Direct, the registration service, will
actually use Kerberos for authentication.  The nameserver is a very
important part of this as it provides the initial set of principle names
and passwords to load into Kerberos.  The qi database and the V5 Kerberos
KDC will be maintained in lockstep.  Eventually only a valid Kerberos
ticket will be accepted for login purposes to the UofI qi server.  The
terminal server software will also be changed in the future to obtain
Kerberos ticket granting tickets to verify users. 

Thanks to Stan Kerr at the Computing & Communications Services Office at U
of Illinois, for providing most of the information in this section, and to
Paul Pomes and Steve Dorner for reviewing it. 

------------------------------

Subject Section 1: Setting up and installing a server