Database Reference
In-Depth Information
10
To Govern or Not to
Govern: Governance in a
Big Data World
The question
of whether Big Data needs to be governed is poised to be-
come a Big Data hot topic (we'd argue that it should have been so from the
beginning). Big Data is a phenomenon and it's changing the data characteris-
tics of
every
system; in order to make Big Data usable, it needs to be governed
such that it's more certain, more trusted. Some believe that Big Data needs to
be analyzed in a pristine form (maintaining fidelity), and that any form of
governance or attempt to “clean it up” might actually throw away some
valuable insight. Others believe that governance capabilities can simply be
“built into” the Big Data ecosystem, if and when you need them. Both of
those views are wrong. The answer to the question posed in this chapter's
title is an unequivocal “
yes
”: Big Data needs to be governed.
It's pretty obvious when you think about it. If data that's stored in tradi-
tional repositories is subjected to governance, the fact that you're bringing in
more of it, or different types of complementary data, or alternative persis-
tence engines, doesn't change anything. Suppose that your client signs up on
your Facebook page as a friend and agrees to share their information with
you; if they later unfriend your business, according to Facebook's terms and
conditions, you are no longer allowed to use that information. Is this built
into your social sentiment governance process? If you're using a test environ-
ment, whose data contains personally identifiable information (PII), it's still
211
