The first step in setting up a new HEPDB server is that of making a home for
it to reside in. In the case of UNIX and VMS platforms this merely
involves the creation of a directory with some arbitrary name. In the following
paragraphs we shall assume the server has been placed in the directory
hepdb/expdb. 
Within this directory one should now create the subdirectories
which are used to hold various data essential to the server. The first of
these should be named logs (this is where the server logs are written), the
second queue (this directory holds new updates from HEPDB clients)
, and
the third save (the place the server saves updates after processing).
In addition, a further directory named bad is used to store files that
cannot be successfully processed.
In the case of VM platforms the following procedure should be adopted. There should be one account per experiment set aside for the server, this should be given a name consisting of the prefix CD followed by the experiment name. For example the CPLEAR collaborations server is named CDCPLEAR. (Note that in the following examples when refering to a VM implementation our example account will be CDEXPDB)
Within this account your 191 addressed disk (this will hold the actual database files) should be allocated some 20 cylinders of space, your 192 addressed disk (the link disk to the HEPDB 191 code) should be given 2 cylinders and finally your 193 mode disk (the journal disk for transactions) should be given 5 cylinders of space.
The final step in placing the server is the setting of an environmental
variable
CDSERV which tells HEPDB where your server resides.
(It is recommended that this assignment be part of your standard setup procedure)
The way in which the variable is set varies from platform to platform, a
list of possible commands follows: