Standardization problem

from Wikipedia, the free encyclopedia

The standardization problem belongs to the research area of business informatics and deals with the question of whether and to what extent components in information systems should be standardized.

backgrounds

In many companies or groups of companies, the application systems have grown historically and in some cases in an uncoordinated manner. This leads to heterogeneous IT landscapes, which complicate the exchange of information between different areas due to incompatibilities .

The use of standards is an essential measure to reduce these integration costs. Standards are either adopted by recognized standardization organizations in a regulated process (in this case one speaks of norms ) or they arise in an unregulated manner due to high prevalence (de facto standards) . Examples include the company-wide use of EDI standards or SOA - platform , which ideally smooth integration of services from different providers based on web service enables standards.

Modeling the standardization problem

The basis of the standardization problem is a graph that consists of nodes. The nodes can be human or machine tasks that store, process and exchange information with one another. There are standards for these tasks. The standardization of a system element causes standardization costs (e.g. for the acquisition of software or the training of human task holders), but simplifies the transfer of information so that so-called information costs can be saved. These consist of communication costs (e.g. costs for the manual processing and transmission of business documents) and friction costs ( opportunity costs of a bad decision, which are based on a lack of information due to the non-standardized transmission). To simplify matters, it is assumed that there is a central decision-making body for all system elements considered.

The system elements are shown as nodes in the model. The standardization of a node causes standardization costs over the entire planning period of .

The decision about the standardization of a node is modeled with the help of the binary action variables .

The edge between two nodes represents the transmission path for information. The information costs that can be saved over the entire planning period if both nodes are standardized are shown at the edge . The information costs between two nodes and are saved precisely when and are standardized, that is . The objective function of the decision problem is then:

example

The trade-off between node-related and edge-related costs can be explained using a very simple 5-node problem with a standard. The numbers in the nodes represent the costs of standardizing the respective nodes. The realizable edge-related cost savings are shown at the edges between two nodes. As already mentioned, these edge-related costs between two nodes can be saved if both are standardized.

In the first figure, no node is standardized, which we indicate by the fact that the nodes are shown in yellow. With this constellation, there are no standardization costs, but all edge-related costs must be borne. This results in total costs of 217 monetary units (MU). In the case of the constellation as shown in the second figure, only node 3 is standardized. This means that standardization costs of 36 MU have to be borne. However, since the use of a standard on just one node still does not support the exchange of information (what good is it if you are the only one to have a fax machine or an e-mail account), the edge-related costs are still 217 GE, which corresponds to total costs in the amount of 253 MU. If node 5 is also standardized, the standardization costs total 61 MUs (third figure). In return, however, the edge-related costs between these two nodes are saved in the amount of 45 MU, which leads to a total cost of 233 MU. If another node is standardized with node 1 (fourth figure), the standardization costs increase by a further 30 MU; instead, the edge-related costs between the three standardized nodes amounting to 100 GE (45 + 35 + 20) are saved. This results in standardization costs of 91 MU and edge-related costs of 117 MU for the constellation shown below on the right, i.e. H. the total cost is 208 MU.

Complexity of the standardization problem

The standardization problem is a combinatorial optimization problem in which the complexity is based on the trade-off between the node-related standardization costs and the edge-related information costs. The problem can be solved with the help of linear integer programming . This solution method becomes problematic in particular when the standardization problem is expanded, since the computing times then increase sharply. If the standardization problem is expanded to include the choice between alternative standards and periods, possible solutions already exist , while the simple standardization problem presented only leads to a complexity of alternative courses of action.

Because of this complexity, scientists in the field of operations research have dealt with the standardization problem and developed alternative methods for finding solutions. Kimms designed a minimum-cut approach which is clearly superior to the originally developed branch-and-bound method in terms of computing time. Domschke / Wagner develop further solution methods and show that problems with one or two standards can still be solved with polynomial effort. However, problems with three or more standards are np-hard.

Areas of application and extensions

Although the standardization problem was originally developed for the field of communication standards, it can be transferred to other domains. Wüstner extends the standardization problem by introducing converters as an alternative to the use of standards. In this way, on the one hand, the opportunity costs can be reduced, which result from not using the standard that best covers the requirements. On the other hand, however, this converter solution usually leads to higher costs than the use of standards, since the integration of functions or areas leads to the need to use up to converters (Wüstner 2005). In addition, the widespread use of such conversion solutions generally reduces the flexibility of the entire IT landscape.

It can be extended to other domains, such as B. apply the use of standard software or the use of services within the framework of service-oriented platforms (Widjaja / Buxmann 2011).

literature

  • P. Buxmann: Standardization of business information systems. Wiesbaden 1996.
  • W Domschke, G. Mayer, B. Wagner: Efficient modeling of decision problems: The example of the standardization problem. In: Journal for Business Administration. 72, 2002, pp. 847-863.
  • W. Domschke, B. Wagner: Models and Methods for Standardization Problems. In: European Journal of Operational Research. 162, 2005, pp. 713-726.
  • A. Kimms: Costing Communication Standards in Information Systems Using a Minimum Cut Approach. In: Journal of the Operational Research Society. 54, 2003, pp. 426-431.
  • T. Widjaja, P. Buxmann: Compatibility of software platforms. In: Journal for Business Administration. 2011.
  • E. Wüstner: Standardization versus conversion: economic evaluation and application using the example of XML / EDI. 2005.