Biomedical Engineering Reference
In-Depth Information
Indexing
Indexing methodology, including selection and use of the most
appropriate controlled vocabulary
Integration
Integration with other databases
Intellectual Property
Ownership of sequence data, images, and other data stored in and
communicated through the database
Interfaces
Connectivity with other databases and applications
Legacy Systems
How to deal with legacy data and databases
Licensing
For vendor-supplied database systems, the most appropriate licensing
arrangement
Life Span
The MTBF for the hardware as well as the likely useful life of the data
Load Testing
The maximum number of simultaneous users that can be supported by
the DBMS
Maintenance
Cost and resource requirements
Media
The most appropriate disk, tape cartridges, and CD-ROM media
Normalization
Avoiding errors by representing data one way, one time, and in one
place
Operating Environment
Ensuring proper power and operating temperature and humidity
Operating System
UNIX, Linux, Windows, MacOS, or mini/mainframe OS
Output
Format of database output
Performance
Access time and data throughput
Privacy
Provision for preserving confidentiality of data
Query Language
Proprietary or standard query language
Redundancy
Hot backups, shadowing, and RAID systems
Resource Requirements
Hardware, software, and operating and development personnel
Scalability
Ability to handle greater data volume with added hardware and/or
software upgrades
Security
Limits on user access, from username-password combinations to
biometrics, as well as encryption of sessions
Stand-Alone vs. Network
And multi- vs. single user
Standards
From media format to operating system, query language, and data
models
Utilities
Availability of software tools for data recovery
Vendor Viability
Commercial viability of the hardware and software vendors supplying
database tools and platform
For example, a milestone in designing and implementing a database is defining the type of data to be
stored. This decision will then imply the most appropriate data model and type of DBMS to employ. If
the data are nucleotide sequences, then a reasonable choice would be a semi-structured database
based on XML-tagged text files. However, if the data are images of 3D protein structures and
keywords, then either an object-oriented or a relational database would likely be more appropriate.
Even though the representation of rows and columns may not be optimum for mapping protein
structures onto a database, factors such as support from a commercial relational database vendor
and support might dictate use of a relational product.
Consider the process involved in creating a central data warehouse of a scale appropriate for the
