The TAL Process
This section covers how the TAL components interact to load data bundles into Servigistics InService.
The TAL process has the following operations:
• Transform – transforms content from bundles into a data format understood by the type definition
• Aggregate – stores the transformed data in the Publication Manager (PBM)
• Load – loads the transformed data into the local or remote viewers (single or multiple)
The transform begins with the Identity Manager (IMAN) component where the latest timestamp for objects is compared against the latest loaded version of the same object. The IMAN database is also a translation dictionary where translation strings, captured from the translation.xml bundle artifact, are kept so that content (such as parts lists) with localized content is brought up-to-date with all translations that currently apply. As the transform progresses, content is output, notifications sent, and eventually a load script generated that maps the generated content into a specific family and segment or segments. This script is used by the Publication Manager load process.
The transform process executes a number of steps in the following sequence:
• Perform Sentinel check – verify the existence of the ReadyForTAL file.
• Check type of bundle – checks for the presence of a manifest *.mf file in the bundle to determine if it is a third party bundle and whether the part relations file exist or not.
• Transform the Px bundle
◦ Pre-transform validation
◦ Initialize the transform process
▪ Create or fetch a unique ID for each object using the ImanManager.identify API. If the object already exists in the Identity Manager, then it provides the last publishing time.
▪ Acquire a lock for publishing the collection. Check if there is any bundle which is already holding the lock on the bundle specific collection. If not, move ahead with TAL.
◦ Validate the bundle
◦ Process <DataType>
▪ Before processing shared content such as Parts and its relations, the transform process tries to acquire a lock on the shared object as well.
▪ Perform transformation on every file present in the bundle.
▪ Save the publishing data of objects through a manifest file into the Transform Registry (TR).
◦ Process - Parts deletion
• Transform third party bundle to a neutral XML representation (Specific to Part and Part Relations)
◦ The third party data comes in CSV format
◦ The CSV data is converted into XML for transform.
◦ The new, updated, and deleted parts are processed using the transform.
• Clean Storages
• Generate notifications
◦ Aggregate notifications created during the transform stages.
◦ Send notifications to configured recipients.
• Purge Exclude Validation
• Optional pause
• Create Transform Completed Sentinel
• Remove Transform Failed Sentinel
The load process first runs Load and Publish Preparation:
• Copy DCTM_output contents to Work folder.
• Remove Bundle based lock.
• Trigger Publish to Preview job.
The Load and Publish Preparation step copies the data from DCTM_Output to the Publication Manager. When Load and Publish to Sites is triggered, the data is loaded into Servigistics InService. This ensures the Publication Manager always contains the latest published content. Post load notifications are also processed and sent to specified recipients.
Here is how the load process works:
1. Publish to Publication Manager
a. Prepare structure of data to be loaded into the system through CMScript.
b. Load into Publication Manager.
c. Prepare packets to be loaded to the viewer system based on data in the viewer.
d. Send the packet to the viewer.
2. Publish to Preview
a. Merge the tables of contents.
b. Index the contents.
c. Process applicability and filtering information.
d. Create and manage links.
e. Update result field specific database tables.
f. Store contents in the repository.