Archiving and Versioning

The Archiving and Versioning features of Analysis Server, added in version 3.0, allow you to publish codes which reside in arbitrary back-end systems, including databases, PDM systems, and Version Control Software. In a rapid prototyping, distributed process integration tool like Analysis Server and ModelCenter users can quickly develop lots of data, wrappers, and models. The Archiving and Versioning features are designed to help alleviate problems associated with this level of productivity with regards to the wrappers and components loaded onto the network with Analysis Server. This provides a number of advantages to systems administrators, model builders, and model users.

Traceability - The back-end system can provide historical information about when and how components published on Analysis Server have changed, including who changed them.
Usability - A Model Builder or Model User has access to a wide variety of codes and versions of codes without having to manually dig old code out of archiving systems and publish them.
Repeatability - A Model User can pull up an old file and re-run it easier since archiving tools typically keep copies of all old files.
Manageability - Management of large Analysis Server codesets can be handled by Version Control Software or PDM systems. Analysis Server becomes an integrated part of a workflow already in place at your company.
Consistency - A network of Analysis Server computers can all publish the same or similar codesets from a centrally managed archive

Virtual Directory Structure and Librarians

The Analysis Server can be configured to use a particular backend database, CM, or PDM tool. It is expected that an administrator may wish to pull from several sources on the same Analysis Server. For this reason, archives may be “mounted” onto logical directories that Analysis Server publishes.

Analysis Server uses pluggable components called "Librarians" which have to ability to talk to specific backend systems. We provide several out of the box solutions and also publish an API which allows you to write custom backend librarians.

Configuring Archiving and Versioning

To turn on Archiving and Versioning in the Analysis Server use the librarianFile option. This will disable the analysesPath , userPath, and smartUserScan options and specify an XML description file which configures your virtual directory structure. An example description file is included in the Analysis Server distribution called librarians.xml.

Comments in the librarians file are started with ''. The file consists of a top level tag '<librarianOptions>' which contains multiple '<librarian>' sections. Each librarian section in turn contains the following fields:

<class> - The full classname of a Java class which implements the IPHXLibrarian interface.
<mountPoint> - The location in the 'virtual directory structure' where this librarian's files are found.
<librarianOptions> - This tag contains further XML whose structure is dependent on the type of librarian described. See the documentation for the particular librarian chosen for a description of the configurable options for that librarian. Note that in the case of Mounter and Combiner librarians, the entire XML structure may be repeated inside of this tag.

You do not need to restart the Analysis Server in order for changes to the librarian file to be noticed. New connections to the Analysis Server will automatically see the changes.