The GroupWise Document Management System

If a GroupWise post office has no libraries, the GWDMS directory will not contain any data. After libraries have been created, though, the structure beneath GWDMS becomes fairly complex.

Shared Document Management Services Database (DMSH)

This file, DMSH.DB, exists at the root of the GWDMS directory structure. It contains lookup tables for document type and other administrator-defined document property fields, as well as a list of the libraries defined under this post office. It does not contain any document data, but it is critical to the functionality of the Document Management System on this post office.

Library Directories and Naming Conventions

Regardless of what you name your library, the library directories created under the GWDMS directory follow a numeric naming convention, beginning with LIB0001 and hexadecimally numbered through LIB00FF.

The last two characters of the directory name become the library number. As you can see in Figure 4.7, this number is used throughout the library directory structure for the naming of files found in library directories.

In Figure 4.7, the last two characters of every file in the LIB0001 directory are 01 (excluding the .DB extensions, of course). Following is a look at each of these databases in turn.

Library Database

DMSD00xx.DB is the library database, in which xx is replaced with the two-digit hexadecimal library number. This database contains the table of document numbers, as well as property-sheet field layout information for this library. It contains no document data but is critical to the function of the library.

Document Properties Database (DMDD)

There are 10 document properties database files; they should be treated as a single database that has been partitioned into 10 files. The naming convention is DMDDppxx.DB, in which pp is replaced by a partition number 00 tHRough 09, and xx is replaced by the library number.

The document properties database contains all information entered into document property sheets. This includes document subject, author, type, associated application, and much more. The property sheet is one of the key benefits to the GroupWise Document Management System, because it allows information about the document to be stored and searched for independently of the document itself, while maintaining a relationship to the document.

Each document record contains three kinds of pointers. The first pointer is to the document BLOB file (discussed later in the section "Document BLOB Files"). The second is to the document's activity log (discussed in the next section). The third pointer points to any user database that contains a document reference to this document. This pointer is used to provide the mailbox indexing process with a word-list file for this document. In this way, a search for full-text of the document returns the document reference, even if the user did not specify a search of GroupWise libraries.

The last digit of the document number indicates in which DMDD partition the document's property sheet record resides. For example, document 1762 in this library is referenced in DMDD0201.DB, and document 21265's properties are in DMDD0501.DB.

Document Logging Database (DMDL)

There are also 10 document logging database files. Again, they should be treated as a single database spread across 10 partitions. The naming convention is DMDLppxx.DB, in which pp is the partition number 00 through 09, and xx is the library number.

The document-logging database contains all activity logs for documents in the GroupWise libraries. Any action that affects a document is logged in this database, including opening, closing, viewing, downloading, uploading, and deleting. Also recorded in the DMDL files is the filename for the document BLOB on which the logged action was performed. When a document is viewed, the BLOB filename does not change, but any action that changes document content (closing it and, hence, checking it back in with possible changes) results in a new BLOB file being created. After the successful creation of such a new BLOB file, the old BLOB file will be automatically deleted, as it is not necessary anymore.

Document BLOB Files

Although the DOCS directory structure is very similar to that of OFFILES, the contents of the 255 DOCS subdirectories are very different from those of the OFFILES BLOBs.

Document BLOB files are created for every document, regardless of document size. Whereas OFFILES BLOBs are created only when field data exceeds 2048 bytes in length, document BLOBs are created even for a 0-byte document.

Word-List Files

There is another kind of BLOB in the DOCS subdirectories. Word-list files are created for each document when that document is indexed. These files contain, as their name suggests, a list of every word in that document. These files are used to build the .INC files (incremental index files), and are also used to provide mailbox indexes with the full-text information for any document references they might contain.

Document Compression

Document BLOB files are encrypted and compressed, just like OFFILES BLOBs, but they contain information that allows them to be manually opened by certain GroupWise users. The GroupWise client can read a BLOB file directly and determine whether the GroupWise user is supposed to have access to this BLOB (a user with the manage right for this library, or a user who has edited the document in the past). If the user is cleared for the document, it will be reimported, and a new BLOB and property sheet created. This might be useful for restoring documents that have been deleted but whose BLOB files exist on backup media.

Document BLOB files are typically around 50% smaller than the documents in their native formats. Pure text documents can be reduced to 10% of their original size, whereas compressed graphic formats (such as GIF or JPEG documents) cannot bereduced.

Document Index

All documents are full-text indexed. This is another selling point of GroupWise document management as a solution for document storage. Documents can be found using the GroupWise client to search for words they contain, or for field data recorded in their property sheets.

Inside the INDEX subdirectory, there are 10 active and permanent index files. These files have an .IDX extension. There are also many .INC files containing the incremental indexes. Just like the mailbox index, the POA QuickFinder Indexing process creates the document index. The .INC files are created from the word-list BLOB files discussed earlier. At the end of the day, all .INC file content is written into the .IDX files, and the .IDX filenames are updated.