Databases Reference
In-Depth Information
income, expenses, cash flow, and the organization's overall net worth. These financial
statements were used to help decision makers think about whether to invest in new
ventures or conserve their cash.
3.2.2
Row stores evolve
Beginning in the 1970s, many companies purchased separate siloed software applica-
tions for different aspects of their business, as seen in the left panel of figure 3.3.
In this stage, one database from a particular vendor might contain human
resource (
HR
) information, another database from a different vendor might store
sales information, and possibly a third vendor system would store customer relation-
ship management (
CRM
) information. This structure mimicked the structure of the
organization, where individual departments operated in their own world with limited
communication between groups. Companies found these siloed systems to be appro-
priate and secure in situations where it was necessary to protect sensitive data (such as
employee salary information or customer payment information). The isolated systems
made it easy for each department to manage and protect their own resources, but it
introduced challenges for the organization.
A key problem with siloed systems was the challenge of creating up-to-date reports
that merged data from multiple systems. For example, a sales report in the sales track-
ing system might be used to generate commission information, but the list of sales
staff and their commission rates might be stored in a separate
HR
database. As each
change is made in the
HR
system, the new data must be moved to the sales tracking
system. The costs of continually moving data between separate siloed systems became
one of the largest budget items in many
IT
departments. To combat the problem of
Siloed systems
ERP drives BI/DW
NoSQL and documents
HR
Finance
ERP
Documents
ERP
Documents
Inventory
Sales
BI/Data
warehouse
BI/Data
warehouse
OLAP
NoSQL
1970s, '80s, '90s
1990s, 2000s
Today
Figure 3.3
Understanding how NoSQL systems fit into the enterprise can be seen by
looking at the three phases of enterprise databases. Initially organizations used
standalone isolated systems (left panel). In time, the need arose to create integrated
reports and the siloed systems were merged into ERP systems with fine-grained security
controls, with separate systems for reporting on historical transactions (middle panel).
These data warehouse and business intelligence systems use online analytical
processing (OLAP) to create ad hoc reports without impacting ERP performance. The
next stage (right panel) is when NoSQL systems are added for specialized tasks that
RDBMSs aren't well suited for, and also serve as a bridge to document integration.
