Biomedical Engineering Reference
In-Depth Information
Figure 2-14. Distributed Database Integration. Distributed databases can
be configured to share data through dedicated, one-to-one custom
interfaces (left) or by writing to a common interface standard (right).
Custom interfaces incur a work penalty on the order of two times the
number of databases that are integrated.
A better solution to integrating incompatible databases is to write interfaces to a common standard.
For example, in clinical medicine, most application vendors are compatible with the Health Level 7
(HL7) interface protocol, which allows Radiology, Laboratory, and Pathology systems to exchange a
subset of their data, such as patient demographics, diagnosis, drug allergies, and current
medications.
Full database integration is much more than simply moving data to a single hard disk. A file server
can store data from dozens of various applications and yet have no integration between applications.
Similarly, just as a single hard disk can be formatted so that it appears as several logical volumes or
drives, a distributed physical architecture can function like a logical centralized database. Taking this
analogy one step further, there are hybrid database architectures that combine aspects of centralized
and distributed architectures to provide enhanced functionality or reduced cost. For example, the
Storage Area Network (SAN) architecture is based on a separate, dedicated, high-speed network that
provides storage under one interface (see Figure 2-15 ). With the appropriate software, a SAN can be
configured to provide the functionality of a central data warehouse, including provision for making
available an unlimited subset of the data from each application database.
Figure 2-15. Storage Area Network Architecture. A SAN is a dedicated
network that connects servers and SAN-compatible storage devices. SAN
devices can be added as needed, within the bandwidth limitations of the
high-speed fiber network.
Search WWH ::




Custom Search