International Journal of Engineering and Applied Sciences (IJEAS)
ISSN: 2394-3661, Volume-2, Issue-3, March 2015
On Control of Data Flows in Heterogeneous
Enterprise Systems
Dusan Pavliak, Robert Vrabel
include transformations, cleaning, derivations, calculations,
Abstract The efficient data integration and the control of aggregations, disaggregation, coding, etc (Fig.1).
data flows leading to formation of coherent integrated database
Data are maintained and transformed in order to meet the
has become a complex problem. The information of the
enterprise include the functions from processing of offers, conditions and structure of data in the target system.
orders via computer design and from production to production
management. Finally, the processed data are archived.
Index Terms system integration, data flow, data import,
heterogeneous enterprise systems, ERP.
I. INTRODUCTION
From birth, a man is continuously forced to process a great
amount of data in his mind. Exchange of the information with
the surroundings allows us to keep our own existence. The
data processing in our mind is permanent , constant but
suggestible. Since the 20th century we have experienced
information explosion. The world is facing the information
overload. However, there are no people who would be able to Figure 1. The transformation and data processing in OLAP
process or even register the information.
The difference between pre-processing and post-processing is
II. CONTROL OF DATA FLOWS that the pre-processing processes the source data into the form
in order to combine the source data with the target data within
Effective data integration and control of data flows leads to the data transformation.
the formation of a coherent integrated data base that is Post-processing carries out data transformation already
becoming more and more complex problem. "Information of combined with the target data. This kind of processing is
the enterprise include the functions from processing of offers, necessary to create derived data values and data calculations
orders via computer design and production to production [4-6].
management. The size of integration is different in various In the next step, target data are loaded from import system.
companies, because business and production processes This step is necessary because, imported data are divided into
require different implementation. Therefore, there are efforts categories according to whether the target records will be
to integrate all these functions in order to increase process added, updated, or deleted.
efficiency and responsiveness. The specialists agree that
computer integration of production systems is relatively
smooth, from technical point of view. The biggest problem is
management of information technology support ,from their
choice to application [1-3]. In other words, the focus of
integration in manufacturing companies is currently changing
from manufacturing systems integration to integration of the
information obtained from consistent data which support and
provide the meaningful use. Control of data flow is the
transfer of data into the import system. Then the data are
loaded into the appropriate data structure of importer storage
interlayer. At this level it is possible to process data with the
use of different ways of pre-processing operations. They may
Dusan Pavliak, Institute of Applied Informatics, Automation and Figure 2. Processing data inside the data
Mechatronics, Faculty of Materials Science and Technology in Trnava,
Slovak University of Technology.
Robert Vrabel, Institute of Applied Informatics, Automation and
This kind of data pre-processing significantly speeds up the
Mechatronics, Faculty of Materials Science and Technology in Trnava, import process, because the data in the source dose, which are
Slovak University of Technology.
79 www.ijeas.org
On Control of Data Flows in Heterogeneous Enterprise Systems
unchanged in comparison with the value of the target
attributes, will be omitted from subsequent processing (Fig.
3).
At the end of the data processing, there is the validation of
process results. If the result of importing is not validated, then
control of the data starts.
Figure 4 . Data mapping feedback
After completing and transferring all data, it is important to
archive the data from import system. Data archiving can
serve for searching import errors or for various reports on the
quality and quantity of integrated data (Fig. 5).
Figure 3 . Data processing in a data warehouse
Then control of the result of data processing will be extended
to control of the amount of changes of individual attributes
and their values. If we can precisely define the criteria of this
control, its parts could be fully automatic. In some cases we Figure 5. Data archiving
cannot avoid human intervention in the process, for example,
when we need to find the people responsible for the source
data and the results of the inspection should be discussed III. CONCLUSION
directly with them. If there are made major changes in the
source system (for example, Bulk update data), the amounts
We were able to increase efficiency and speed in production
of imported attributes can be unexpectedly increased, but
by error control followed by the data flow adjustment before
even the increased amount is shown to be valid for the import
processing in the system.
[5].
Parameters for validation are set for each import in advance REFERENCES
and they prevent damage of the target data. For example, they [1] CECELJA, F. Manufacturing Information and Data Systems: Analysis,
include the boundary conditions for a number of processed Design and Practice (Manufacturing Engineering Modular Series).
Elsevier Science & Technology, 2002, 160 strn, ISBN: 1857180313
records, or define valid values of attributes. If the results [2] CUMMINS, F. A. Enterprise Integration: An Architecture for Enterprise
contain satisfying criteria, integration application can be Application and Systems Integration. John Wiley & Sons, 2002, 468
started. The data that are identified as acceptable for further strn, ISBN: 9780471400103
processing, are transferred to integration application. There is [3] VERNADAT, F. B. Enterprise Modeling and Integration: Principles and
Applications. Chapman & Hall, London, 1996, 320 strn, ISBN-10:
the transfer of individual records into the target system and 0412605503
their updating. The feedback, implemented within integration [4] SHERIF, M. H. Handbook of Enterprise Integration. Auerbach
application can control values of individual attributes in the Publications, 2010, 728 strn, ISBN: 9781420078213
target system during the actual import. Data are imported into [5] LAM, W. ,SHANKARARAMAN, V. Enterprise Architecture and
Integration: Methods, Implementation and Technologies. IGI Global,
ERP system, which serves for operational activities (Fig. 4).
2007, 364 strn, ISBN: 9781591408871
[6] GOLD-BERNSTEIN, B., RUH, W. Enterprise Integration: The Essential
The feedback enables us to eliminate the errors caused by the Guide to Integration Solutions. Addison-Wesley Professional, 2005,
change of data in the target system during import processing. 432 strn, ISBN 10: 032122390X
If the system finds such records, that have changed values,
Dusan Pavliak is studying a doctorate degree at Slovak University of
and they do not meet the import conditions, this kind of Technology. His research is focused on data mining techniques.
information is recorded in the importer log structure. If the Robert Vrabel was born in 1967. As a associate professor of Slovak
data satisfy the conditions, they are successfully imported into University of Technology, his interests are in the theory of dynamical
the target system and they are also recorded. The result of systems and their modeling and simulation.
import is first recorded in the integration module of the target
application and then all results are transferred back to the
importer structure of data flow.
80 www.ijeas.org
