by Janette B.
Bradley
axsWave Software, Inc.™
This is an edited version of a presentation I gave at the American Society for Information Science Annual Conference (October, 1996). The presentation was in two parts. The second part offers a short list of tips to help you design easier to use web-query forms.
The Web seems to offer a significant opportunity to distribute data to organizations and corporations.
Due to:
- Existing infrastructure
- Lower cost per station than proprietary/custom LANs
- Seemingly quick deployment of existing databases using HTML and freely available CGI scripts.
A Database is an information or data store accessed through a query language or Application Programming Interface (API)
A Web-based database fits this description with some important differences:
- Access is through Web applications, usually a form, which means that queries must be reformulated to SQL.
- Access interfaces are multi-platform. Often, designers will have no clue what hardware or software will be used.
- Access is often global, which means that profiling the user is far more difficult.
The Relational database model is:
Made up of tables
- consisting of rows and columns
- each column has a name and single data type
- multiple dimensions are represented by multiple tables which are joined to construct a multi-dimensional object or by multiple rows in a table
Relational databases support
- integers
- floating point
- character strings
- date-time
- currency
If it doesn't fit any of the above, then it is a
- BLOB - binary large object.
Because object-oriented databases are good at handling BLOBs, and the new world of information is all about BLOBs.
BLOBs like:
- Images
- Video
- Audio
- Animations
- Mixed Media
An Object Oriented Database:
- Provides for data objects that store in one data structure:
- The data
- Rules and ways about how the data looks, acts, how it can be manipulated, and how it can be accessed.
- Allows for complex data types to be intermingled in one database.
- Allows for algorithms that let the database search in multiple dimensions.
In an OOD environment, developers can create structures that model how the data will be used.
An OOD and its database management system (DBMS) is aware of how to:
- Access or extract internal components of an object. For example, one or two frames of a video.
- Execute operations or functions against objects without exporting them to the client.
- Extract enough about the object to develop an "intelligent" search plan to optimize performance. For example: The user wants multiple frames of a video, plus info on actors, royalties and rights. The OODBMS gauges the speed of retrieval for each item and optimizes a retrieval plan using SERVER resources, freeing the client to continue work.
RDs:
- Mature
- Extensively tested
- Vast amounts of data in this format already
- Programmers know how to optimize for high-speed retrieval
OODs:
- New
- here is a general shortage of experienced, quality programmers
- Lack consensus on standards, definitions, etc.
- Performance concerns
Try to combine the best of both worlds
Maybe?
- OOD evangelists claim that RD/OOD models have the same inherent problems representing complex data that RDs have.
- RD/OOD adherents claim that this model provides an effective bridge to legacy databases and has less performance and reliability problems.
Feel free to contact me with questions or to talk about how the axsWave Software team can help you with your Web site:
Janette B. Bradley (janette@metronet.com)