Download as pdf or txt
Download as pdf or txt
You are on page 1of 10

SYSTEM DESIGN CONSIDERATIONS* The steps in the physical design of a system are as follows: - Input to the system - Output

from the system - Files to contain the operational data - Files maintenance procedures - Programs and job control language systems - Control and security.

OUTPUT DESIGN: An output is essentially specified by what it contains. The contents are decided on the basis of requirement or purpose to be served. The contents refer to data items that will appear in the output. Once the contents are finalized, then the analyst must design the format in which the contents will be obtained. The format is important where the output is for persons. It should include titles, explanation/explanatory comments, and etc. irrespective pleasing get up. The format may also depend upon the media on which the output is to be produced e.g. line printers, DRT terminals, plotters, microfilms etc. Since it is the output through which a system meets its requirements and objectives. It also allows the user to evaluate system's usefulness. The important considerations in the design of output are- usage, contents, aesthetics (the science of beauty in art and nature). The output may be classified as: - External- that goes out of organization like invoice. - Internal- that are used in the organization itself, either for decision-making or for administration of functions. - Governmental. The outputs can also be classified according to the characteristics of their contents: Detail reports- contain data of all entities or transactions. Exception reports- reports prepared during unusual situation (predefined interms of certain condition of data). Periodic reports- produced at regular intervals Adhoc reports- produced on request. Inquiry reports- produced for inquiries (such as writing of new programs or an information retrieval system).

Also, outputs can be obtained from batched systems or from real-time system outputs.

Finally an output may or may not re-enter the system as input, when it happens then there may be following two situations: Turn around documents- a document is produced as output, sent to user, user fills some parts of it and returns the document for input. Output not for users, but for subsequent input only. Here we can consider the output as file.

The format design or form design considerations are similar, whether for output/input. This will be discussed in the next section. But, there are certain additional considerations: - Use proper heading for identification of output and its purpose. - Perform editing on values before outputting them. - Expand internal codes into understandable word. - Include date of production and page number. - Include explanatory comments for interpretation of output. - Use special symbols or underline to draw attention to significant data. Finally, the analyst must specify how the output is to be disposed after use: - to file (how long?) or destroy. This is important if the contents are of confidential nature.

INPUT DESIGN: After completing the output design, the analyst can determine the inputs that will be required for producing those outputs. Moving from outputs to inputs thus automatically establish the relevance of each input entering the system, as also the type of processing it undergoes. Some of the input for producing an output may come from files. Here, we will discuss only the input entering from outside the system. The inputs can also be classified by their important characteristics. The inputs may be generated within the organization or outside it; the processing of inputs may be batched or on-line, etc. Whatever may be the nature, the following activities are associated with each input: Recording of the data on proper medium. Transmission to the place of processing. Transcription to the computer-acceptable medium. Verification, sorting and controls for ensuring complete and correct processing.

Some of the cost is associated with above steps, and the analyst should consider possibilities of reducing or eliminating these to the extent possible. The design of input basically consists of defining contents, selecting the medium and deciding the format in which the contents will be entered. To carry out these effectively, the analyst must recognize the following important characteristics: - Identification of the input - Frequency with which it is generated. - Volume, both average and peak. - Validation procedure required. In designing the contents, i.e., the data items that appear in the input, following considerations are important: - The data items should be meaningfully related to each other. - They should be available at a single place, unless, it is required. - The contents should include some redundancy for applying validation checks. - The format of data values should confirm their processing. Next, the analyst decides the medium on which input will be obtained. The important factors in selecting media are: - Method of processing (batch or on-line) - Device available - Environment or operating conditions prevailing at the place of data capture. - Volume of data and cost of data preparation all-inclusive. - Medium re-usability.

The forms design is concerned with the layout. Proper spacing provided. The hand filling should be minimum. The important immediately. Windows should be prepared for values that are values should appear in a logical sequence, e.g. addresses in numbers and dates at the top right, signatures at bottom, etc.

and margins should be data should be located to be transcribed. The the top left, sequence

Finally, the analyst must develop criteria for validation by inputs, viz. Type check: numeric or not Limit check Compatibility checks between two or more items

FILE DESIGN The important steps in the design of files are as follows: * Understand clearly the inter-relationships between data elements that are to be put in files. This understanding should be over all data elements. Identify entities, attributes that identify and characterize the entities, relationships between entities, and attributes of these relationships. * Make an in-depth study of how the operational data will be sued by processes. For each processing task, determine:

Frequency of accessing Criteria, by which, data is selected for processing. Volume or percentage of total volume used in each execution of the task. Data elements used in the task, and The sequence, if any in which the task requires input and produces its output.

* Determine volatility (i.e. update characteristics and growth characteristics of operational data, total volume of data. * Determine the content of each file. The main aim should be given to the relation of data elements and data usage. * Select file organizations, accessing methods, processing aspects (e.g. buffering), and decide protection and access controls (e.g. labels) * Design security mechanisms to prevent loss of data and/or processing due to system crashes, program malfunctions, and accidental or intentional destruction of data.

The primary objectives in file design are: To maximize efficiency of data processing to minimize storage and accessing costs. To minimize the complexity of the resulting system, i.e., it should be easy to understand, operate and maintain, and To minimize complexity of programming.

E.g., fixed length record files are easy to program than varying length record files; sequential processing is usually more complex than direct accessing; retrieving required data distributed in two or more files is complex compared to where it is available in the single file, etc.

The physical file design considers usage of data to decide file contents, as they will appear in actual implementation. The normalized files are inputs to this step. The following actions are usually carried out in physical design: Merge two or more normalized files into a single physical file because they are generally used together during processing. This merging will reduce the number of files and make processing efficient. Split a normalized file into two or more files because whole of its record contents are never or very infrequently used together in processing.

DATA BASE The specific objectives of database design are: (1) Controlled redundancy: Redundant data occupies space, which is wasteful. If versions of same data are in different phases of updating, the system often gives conflicting information. A unique aspect of database design is storing data only once, which controls redundancy and improves system performance. (2) Ease of use (3) Data Independence: An important data base objective is changing hardware and storage procedures or adding a new data without having to rewrite application programs. The database should be "tunable" to improve performance without rewriting programs. (4) More Information at low cost: Using, storing, and modifying data at low cost are important. Although, hardware prices are falling, s/w and programming costs are on the rise. This means that programming and software enhancements should be kept simple and easy to update. (5) Accuracy and Integrity: The accuracy of a database ensures that data quality and contents remain constant. Integrity controls detect data inaccuracies where they occur. (6) Recovery from failure. (7) Privacy and security: For data to remain private security measure must be taken to prevent unauthorized access. (8) Performance: This objective emphasizes response time to inquiries suitable to the use of the data.

FILE ORGANIZATION File is organized to ensure that records are available for processing. The file should be designed in line with the activity and volatility of the information and the nature of the storage media and devices. Other considerations are: 1. Cost of file media (highest for disk, lowest for tape) 2. Inquiry requirements (real-time vs. batch processing), and 3. File privacy, integrity, security and confidentiality. There are four methods of organizing files: sequential, indexed-sequential, inverted list and direct access. A comparison between advantages and disadvantages of file organization is given in the table below: Table: File organization methods A summary Method Sequential Advantages Simple to design Easy to program Variable length & blocked records available Best use of storage space Records can be inserted or updated in middle of file. Processing may be carried out sequentially or randomly. Used in application requesting specific data on multiple keys. Records can be inserted or updated in middle of file. Better control over record allocation. Disadvantages Records cannot be added to middle of file.

Indexed-Sequential

Unique keys required. Occasionally slow Periodic reorganization of file is required.

Inverted list

Random

Calculating address required for processing. Variable-length records nearly impossible to process.

PROGRAM DESIGN It is concerned with defining units of work for the computer. One program is viewed as one unit. During the analysis phase, the analyst identified the tasks that are to be performed by computer. These tasks will generally be of the following types: - Validation and editing of inputs. - File maintenance - Output generation The tasks have to be performed in same sequence, i.e., they have predecessor and successor relationships among themselves. The other characteristics of tasks will be inputs used and files referenced. A program may be defined for each task or program may incorporate two or more tasks or complex tasks can be broken into two or more programs. The program design is basically concerned with deciding allocation of tasks to programs. For this following considerations are important: - A program's resources requirements should not exceed those available with a computer. (The computer system will constrain the number of devices for holding files, the total memory available) - The tasks combined in a single program should be functionally related to each other, so that the program also serves a unique function of the organization. - Combining two tasks, having same input or referencing the same file in similar fashion, into a single program thereby increasing the system efficiency. - Combining tasks increases resource requirements of a program (more memory, more files, etc.) and makes changes difficult. - Lesser the number of total programs in the system, more easy is the system to operate. - Tasks with different frequencies of execution should not be combined into a program. - A complex/large program is difficult to develop, test and maintain. At the end of program design, the task of developing programs is assigned to programs. The programs should be given complete specifications for each program so that he can get complete understanding of his work. He then develops detailed structure of the program and code, implements and tests it. As the programmer develops the logic of a program, certain checks may be used to see completeness. These checks are (common to most programs): Initialization for variable and loops. Loop control and exits Input/output functions and input validations Last record processing Error routines Heading outputs, etc.

The programmer should also give due consideration to the efficiency of a program, especially when it is a time critical (e.g., in. on-line system). The data types, data sizes, data representations, statements used, and idiosyncrasies of the available compiler all contribute to the efficiency factor.

CONTROLS AND SECURITY The reasons of building control and security measures into a computerized system are: - To ensure that it meets the objectives it is designed for. - To ensure that it is correctly operated. - To safeguard the system against error in input, output, etc. and - To protect the system against unintentional or deliberate mischief. If these measures are built into the system during design stages itself, an uninterrupted and correct flow of work is ensured. Following are the considered control measures: (1) Management Controls: The system (data processing) should be useful to the management for the control of organization. The management must be able to plan, to set standards, and to measure activities, effectiveness and performance by using the reports of the system. E.g., the management wants that the organizational units stick to budgets; the system must report significant deviations from the budget so that the management can take corrective actions. E.g., a system designed for keeping tight control over inventories. (2) System Controls: To ensure smooth working of the system, the controls can be of the following types: * Operations control- consisting of feedback about how the system is functioning. The feedback system information consists of: Log books, tickets indicating batches of inputs received, etc. Messages obtained during program runs Control values (records processed, etc.) User and customer complaints Error analysis reports, etc. * Input Controls- like checks on data entering the system. * Output controlsProper scheduling Keeping records/logs of output generated. Reporting exceptions during program runs. Control totals, sequence checks, etc. Directives for output. * Procedure control- Identification of running program on the log. - Restart and checkpoint considerations - Beginning and ending messages and control totals and figures. - Label verifications, etc.

(3) Audit Control: This is to know what has been processed by the system and how. It is generally practiced for management to appoint auditors to examine accuracy and efficiency of the system and its controls. A computerized system must be auditable, i.e., by maintaining logs containing trails of transactions and actions (called audit trails) and keeping files which contain changes to data files, the system should enable auditors to trace data and processing. Another advantage of keeping trails and journal is that recover can be made from errors and disasters. (4) Access Control and Security: The system may contain sensitive data, access to which should be controlled by defining user categories and passwords. The access control has to be planned within the facilities provided by the operating system of the computer. Another aspect of security is to recover from errors and failures. Towards this purpose, the analyst should define backups for data files (which are to be created periodically, and transaction logs and change journals).

This document is copyright protected. Do not circulate it without taking permission from Dr. Alok Bansal (alokbansal75@yahoo.com).

You might also like