Page Comparison

A data link lets you feed data into a workbook without using an import job. The difference between the two job types is when the full data set is loaded. A data link fetches only the preview data for the workbook view; when you run the workbook, the full data set is used. When you run an import job, the full data set gets loaded. You can create a data link from an existing connection or by selecting a new connection. See Types of Data Supported for information about the types of files that Datameer X can be linked to.

You can edit, rename, create a copy, run, view details and information, view the full data, or delete an existing data link.

...

1. Click the + (plus) button and select "Data Link" or right-click in the browser and use the context menu to select "New Data Link".
   
2. Click "Select Connection" to select the connection and or "New Connection" to add a new connection if needed. 
3. Specify the File Type. See the sections that follow for additional details about importing each of the file types. 
   
   • Apache log: Specify the file or folder and the log format. See the samples provided in the dialog box for details.
   • CSV/TSV files: Specify the delimiter such as"\t" for tab, comma ",", or semicolon ";", specify whether the first row contains the column headers and click Advanced Settings. In Advanced Settings, specify the escape character to "escape" processing that character and just show it, set the quote character, and if enter strict quoting is checked, characters outside the quotes are ignored.
   • Fixed width: Specify the file or folder and specify whether the first row contains the column headers.
   • Mbox: Specify the file or folder. This is a format used for collections of electronic mail messages.

   • Text files: Specify the file or folder, a regex pattern for processing the data, and specify whether the first row contains the column headers.

   • Twitter data: Specify the file or folder.
4. Enter file and folder information, schema detection, date and time filters, time partitions (see below), or advanced controls in the Data Details tab and click "Next". 
5. View a sample of the data set to confirm this is the data source you want to use, and mark the checkboxes to select which fields to link into Datameer. 
   
    You can specify how to handle empty fields and invalid data. The Empty value placeholders section gives you the ability to assign specific values as NULL. Values added here are not imported into Datameer.
   
6. Click "Next".
7. Define the schedule details and click "Next". INFO: See Configuring Workbook Settings for information on the schedule details.

8. Anchor
   sample_data sample_data

   Enter a description and select how to handle sample data.
   You have the option to Generate Sample data immediately after saving the data link (this saves the sample data to the DM private folder during the data link ingestion) or to Defer Sample generation to workbook, which defers sample generation until using the data link in a workbook. Deferring sample generation is useful for when the data link is available to multiple users but each user may have different access to the original source of the data link. When deferring sample data, the sample data is created and stored in the DM private folder by the workbook user rather than the data link creator. This sample data access still follows impersonation security policies (e.g., Sentry/Ranger). 
   Click "Save".
   
9.  Name the file, and click "Save".

...

1. Create a connection using Hive.
2. Create a new data link and select your Hive connection.
3. Select which tables and databases you want to ingest on the Data Details page.

4. On the Filter page select Enable partitioning. 

   Note
   As of Datameer X 7.4 Performance and speed have been improved when filters set on a partitioned workbook column pushes from your Hive data link. Before, the full data in the Hive table was being read and filtered which included your column's partition. Now, only the selected column's partition is being read and filtered.

5. Define the date pattern to map to Hive's time based partitions. For example, if in Hive you have three columns with integer values which contain year, month, and date, then each value must be defined in a different field.

   

   If you have one column with a string value, use one field to define the pattern, such as yyyy-MM-dd. 

   

   In the Advanced section you can select a check box to set how to handle invalid partition values. If you check box selected, partitions that don't match the pattern are ignored and the data link job runs, logging ignored partitions in the logfile. If the box is not selected, the data link job fails if it encounters a value that does not match the partition pattern.
   
   As with any other partitioned data, when creating or opening the linked data in a workbook the partitioned data is displayed. You can then select the data you want to analyze.
   

...

The delete feature deletes the data link in Datameer X but doesn't delete the actual data.

...