Graphics Reference
In-Depth Information
Chapter 14
Big Data
Big Data
has been at the center of much of the innovative research and
development in the field of graph analytics and visualization. With so many
of our social, consumer, and other exchanges occurring online, the amount
of data being collected every day, as well as opportunities to link across
data sets, stretches the limits of our ability to take full advantage of it. For
businesses, the central problem is no longer getting better data, but
getting
better information out of data
.
The term “Big Data” can mean different things to different people, but
defining issues are generally agreed to include the four V's—volume, velocity,
variety, and veracity. Simply put, challenges exist with the size of data, how
rapidly it is streaming in, how extremely multi-faceted it has become, and
how uncertain some of the source or derived data can be. Big Data is not
strictly defined by how big it is, but by the fact that it is large and complex
enough that it defies management and analysis using traditional systems and
approaches.
Traditional systems often store structured data in table form on a server.
Queries are then used to slice and dice along dimensions for subsequent
analysis. Analysis is done in the memory of a single machine, typically with
facets displayed independently in separate views. In some of the more
advanced tools, filtering and interactive cross-view highlighting provide the
capability to explore relationships one at a time with dimensions that are
outside the view of a single facet.
By contrast, Big Data is often rightly associated with alternative data
structures and distributed systems, where processing and management tasks
are spread across a cluster of computers. In reality, however, an extremely
wide variety of technologies are involved that may, at times, include
traditional systems such as relational databases. Because the data is often
complex, so are the back-end processes and systems for exploiting it.
Big, complex data also tends to defy traditional approaches to visualization
and analysis. Slice-and-dice approaches, though often useful for
understanding overall characteristics of data and discovering broad patterns
across one or two dimensions, provide a view of only one facet at a time.
