Once the directory synchronization system has been analyzed, the various steps that the directory data have to go through can be planned. This involves assigning a unique name to each directory, designing the cooking and serving recipes for each directory, and determining where each step is to run.

Each directory needs a unique name. This name is used by the various DIRBOTs to determine what processing the directory needs, and what to do with the resulting data. This should be a short but descriptive name.

The syntax and semantics of the recipes are discussed elsewhere, but it is worth noting that there is a stand-alone utility which can be used to verify recipes. For each directory agent, there are also utilities which interact with the directories just like the directory agent master and slave channel programs, but which use local files instead of receiving or sending directories by e-mail. This enables you to extract sample LDIF files to test cooking recipes. There is also a stand-alone cooker which allows you to cook such an LDIF file producing a cooked LDIF file. The serve process can be tested by running the stand alone cook utility using a serve recipe to process a cooked LDIF file: the result should be the file you started with.

Note that it is important that the cooking and serving processes are performed by a DIRBOT configured on the same system where the e-mail channel associated with the directory runs. This is so that any e-mail address manipulation is performed with the same context.

The directory differencing process can be performed by any suitable PMDF system. This might be the same system as the one which does the cooking and serving of a directory, or it might be one which is otherwise uninvolved in the directory synchronization process.