Database Reference
In-Depth Information
The types of data stored in tables are usually more organized physically than
in spreadsheets or other locations, but this can be deceiving. Just because
the data is organized well physically, stored in tables with relationships, and
labeled in detail, doesn't mean that the application or user process that is
populating them has the right amount of governance or access rights built in.
That governance is critically important because when you are building data
sources for visualization, data quality is so important. A key part of creating
data visualizations is knowing the data well enough to help others understand
and interpret it more effectively. Storing data in tables is normally a byproduct
of that data coming from another source where it was worked with previously,
then put in a table, and is now persisted that way. If you're going to keep the
data, add to it, and continue to improve it, a table is often a better long-term
location than a spreadsheet that could be easily deleted from the file system.
oLAP AND TABuLAr MoDELS
OLAP
and
tabular models
are terms for systems that store data in a special
format called multidimensional, or in the case of tabular, as a column store
with a multidimensional interface. This means the data is designed and stored
in a way that allows the end user to connect data across different business
segments more easily. These models are usually set up by developers or
power users and deployed to a server in your environment. This provides
a trove of data for end users to navigate through and begin to explore and
build visualizations. These are very common data sources and are made up
of data from tables and other sources to consolidate the information in one
place that other tools can connect to and analyze.
PowerPivot is an instance of a tabular model embedded inside Excel and can
be published from PowerPivot to Analysis Services.
rEPorTS AND DATA FEEDS
Reports and data feeds are also good sources of data. May tools these days can
connect directly to reports and data “feeds,” or online streams of data. These
tools can pull this data in as if it were a spreadsheet or table and enable users
to work with it in more creative ways. These sources are advantageous because
many times the security and integrity of the data was already addressed
upstream in the process. The risk is that using reports as data sources for other
reports can lead to a long chain of reports pulling data from reports and lead
to a scenario where the data has gotten polluted—that is, the figures coming
through are wrong.
