3 External data loads

• Set up file watchers and load data to Development and QC before you promote
  
• Create input data models for new data sources by loading an externally-created MDD file
  
• Accurately define data sources
  

Set up file watchers and load data to Development and QC before you promote

• Recommendation: Follow the workflow of loading data to the Development and QC lifecycles before you promote to Production.

• Rationale: Loading data into all source models and performing a fully qualified system test in the QC lifecycle is the most effective way to determine whether a study is ready for Production.

• Additional information: The QC lifecycle is meant to be used as a validation lifecycle, and allows you to verify working objects beyond initial setup within the Development lifecycle.

Create input data models for new data sources by loading an externally-created MDD file

• Recommendation: When you add a data source whose metadata hasn't previously been used in a specified DMW environment, create the source data model by loading an externally-created MDD file.

• Rationale: It's quicker and less error prone to define columns and other data model properties in an MDD. In addition, it's relatively easy to modify and re-load an existing MDD file if you need to make changes.

• Additional information: Oracle Health Sciences Consulting has worked with customers to develop several different MDD generation macros and utilities to further increase the ease of working with MDDs outside of DMW.

Accurately define data sources

• Recommendation: Make sure to define data sources and associate them with the correct vendor.

• Rationale: Defining data sources for file-watched data, and assigning the correct vendor to the files they provided improves DMW's filtering functionality discrepancy handling.

• Additional information: You can select discrepancies with a common data source, and select Send to Spreadsheet to produce a file that lists the discrepancies for a single vendor. Properly defining data sources for input data models prevents cross-contamination of data issues from different vendors and the manual effort that is required to distribute discrepancies to the correct recipients.