Database Reference
In-Depth Information
There is a downside to performing the work on the client—it usually means
that you need the data to exist on the client as well. For big datasets,
requiring client-side data can mean spending a lot of time reading your data
(especially if that data is stored in the cloud).
When interacting with BigQuery (and other Big Data sources), however,
Tableau takes a hybrid approach: It issues BigQuery queries to perform
aggregations and then downloads the results of those queries locally. This
means that only the query results need to be transferred to the client, and it
doesn't need to rerun the BigQuery queries unless the aggregations change.
In this model, many data manipulations can be done with only local data.
There are a couple of potential “gotchas” when using BigQuery with Tableau.
The first is that you don't have much control over the SQL that is generated.
Tableau currently uses the Simba BigQuery ODBC driver to issue standard
ODBC requests (as Standard SQL). This doesn't allow usage of specialized
BigQuery SQL syntax, such as using the
EACH
keyword to group by a field
that has a lot of distinct values. Tableau treats BigQuery like a standard data
source, so it doesn't have many specializations that would make BigQuery
perform better.
It is possible that by the time you read this, Tableau will have released a
custom connector that can more effectively handle BigQuery data sources.
Tableau has specialized data connectors for other data sources (for Apache
Hive, for example). A specialized connector would mean that you wouldn't
need the Simba ODBC driver; you could run in environments in which the
ODBC driver isn't available (such as on a Mac).
Tableau has a free 14-day trial that you can use to try out the BigQuery
integration. To sign up, just navigate to
http://www.tableausoftware.com/
in your web browser and click
the Free Trial button. Select the Tableau Desktop option and follow
instructions from there to download and install the software.
