Friday, October 3, 2014

Force Extract To Be Built On Server Instead Of Desktop

The reason for using this trick would be if you have a large dataset you need to extract and want to prevent the extract file from being built on your desktop before you publish to Tableau Server.  Using this technique we can build an empty extract with 0 rows of data in it on your desktop, and then as soon as you run the extract refresh on Tableau Server (either manually using 'Run Now' or on a schedule) it will populate with the full dataset.

First step is create a calculated field which has the formula:
datetrunc('minute', NOW())


































This will roll the current time back the beginning of the minute (which makes it important that you do the rest of this quickly).

Next right-click you data source and choose 'Extract Data'.  Inside that window choose to add a filter and pick the new calculation you just made:




























On the next window pick 'Individual dates and times':




















And then 'exclude' the single timestamp you see in there:
























Your extract window should look like this and then you can click 'Extract':

























This should build an extract file that has 0 rows of data in it.  You can see this by right-clicking on the data source and choosing 'Extract->History':




















See that it imported 0 rows (my timestamp is slightly different than before because I had to take new screenshots):













Note:  If it didnt work it is probably because the minute has changed and the timestamp we excluded is too old.  If this is the case, change back to a live connection, then right-click the datasource and choose 'Refresh' (or click F5) to dump the cache and send a new query.  This will also update the timestamp to the current time and you can try again.

Now after you publish your workbook and either run the extract refresh schedule, or just pick 'Run Now', it will populate the full extract:











Yay our workbook now has data in it and we never had to build the giant extract on our desktops before publishing!









-The Don Data

1 comment: