21 - Data Processing Architectures

You would use remote job entry when the cost of storing data on the host is expensive or when the user needs to do some local processing and preparation of the data before sending it to the host. This is particularly true when the cost of processing time on the host is expensive and the user wants to do local processing for as many tasks as possible. Users may also feel that local data storage is more secure. Users may also not have access to the processing software locally and need to run their programs on the host because that is where the software resides. The host may also have large amounts of memory or processing power (supercomputer) that the user’s program requires.

There is a danger of repetition and redundancy, the same application may be developed and the same data may be stored in many different systems. The data on the systems are also not readily shareable. The data may not be as reliable and may not be updated as often as desirable. The databases are not always secure nor are they always backed up. When the user is no longer in charge of the database, the application and data are often lost because they are not documented or the organization does not realize they exist. Finally, some individuals may spend more time fiddling with the database and not getting their assigned work done.

When a firm moves from a centralized database to a distributed database it may lower its communication costs. This happens because the data that are most frequently accessed by a specific location are located at that location; thus there are lower communication costs to retrieve it. A distributed database is more complex than a centralized one and may be more costly to develop and maintain.

Transparency in the context of a hybrid architecture is that the user sees the database system as a seamless system that readily provides needed information. The location of the data, the storage format, and the access method should be invisible to the user.

Fragmentation independence means that any table can be broken into fragments and then stored in separate locations. The table should be able to be broken down by column or by row.

DBMS independence refers to the ability to set an organizational standard for DBMSs and allow the local sites to utilize whatever DBMS they want as long as it confirms to the organizational standard. For instance, the standard may be a relational database that uses standard SQL in which case several different DBMSs can be used that fit this requirement.

ODBC (Open database connectivity) allows different applications to talk across platforms and servers via ODBC database drivers. By using ODBC drivers, an IS department can link together a wide range of databases, thus supporting hybrid architectures.

Remote job entry allows the professor to manage her data locally while executing the complex model remotely.

For the company to effectively create a global database they will probably need to design a distributed database utilizing client/server technology. Because each country has its own database, it is probable that the database can be fragmented in such a way that the data most often used by each country will reside within that country. The key factors include knowing how the data are used, what kind of database is currently in place, and the independence of the separate databases in terms of the hardware, operating system, network, and DBMS. Problems to anticipate include:

• The inability to integrate the current systems
• Political problems which may include organizations that do not want to share their data and organizations that do not want to take the time to create a global system
• Data integrity
• Communications infrastructure problems (continuous operations)
• Potential legal problems associated with where the data can reside.

Students should investigate the options offered by firms such as Amazon web services, Google Cloud Platform, and Microsoft Azure. Because of the changing nature of these offerings, the answer keeps changing, but students should consider the ability to create a short-term cluster, ease of software installation, ease of access, ease of use by students, and cost.