Monday, October 27, 2014 6:40:04 PM
I have been asked to find data in a Teradata data warehouse without being provided any contextual 'sherpa guide'. Fortunately, the requested person is familiar with the data that they seek and already has sourced a vetted version of the data through other means. When that person then wants the data to come from a Teradata connection, it is typically because the vetted source doesn't have a convienent means to enable Tableau to connect to the data directly, but the requesting party knows that the data resides in the data warehouse and has obtained proper permissions to read the data directly.
After securing permissions and having knowledge of what the 'right' data should look like, I then connect to Teradata only to find something like 1 million unique Database/Table/Column combinations. To assist my finding the data, I need to create my own sherpa guide. Fortunately, DB's that run Teradata DW typically have instilled common naming conventions to their databases/tables/columns and rarely have obfuscated those items. Connecting Tableau to the DBC database and the columsv view provides a comprehensive list of databases, tables and columns that then can be filtered using wildcard matching to find potential data sources for the requested data.
After finding potentials, creating a connection to each found db/table enables the data discovery (and lets you know if you have permissions to see that particular db/table/column). The last step is to verify the quality of the data to the known vetted source. With the last step complete, going further to actually create data visualization stories with Tableau can begin.