There can be no doubt that the architecture for analytics has evolved
over its 25-30 year history. Many recent innovations have had significant
impacts on this architecture since the simple concept of a single
repository of data called a data warehouse. First, the data warehouse
appliance (DWA), along with the advent of the NoSQL revolution, selfservice
analytics, and other trends, has had a dramatic impact on the
traditional architecture. Second, the emergence of data science, realtime
operational analytics, and self-service demands has certainly had
a substantial effect on the analytical architecture.

The single data warehouse repository simply could not support any and
all analytics anymore. A new architectural concept called the
Extended Data Warehouse architecture (XDW) has taken the place of
the single repository idea. It accommodates the new forms and
volumes of data, the need for different sub-environments for varying
analytical requirements, and the immensely innovative technologies
available today.

In this paper, we focus on the DWA and how it has evolved over the
years since its introduction. The XDW architecture is then described, in
which the need to maintain the data warehouse is documented while
adding new components and capabilities to extend the analytical
capabilities. This section also discusses the appropriate usage of
appliances within the XDW. The rest of the paper covers the benefits
from implementing the DWA, the selection considerations for them and
what the future holds for them.