purposes. Finally, Response Time have multiple
negative contributions since the duration for the
input file download must be minimized as possible
in order to accomplish the real-time constraint that
data must not take more than five minutes from the
moment the download starts (FR responsibility) until
data it made available to the user in the client tools.
Four concerns have been identified as the most
required within the File Retriever system, namely:
Configurability, Bandwidth Usage, Data Recovery
and Error Notification.
The Configurability concern is required by many
FR’s operations since all download activity is
configured, or based, on metadata (e.g. the sites for
the data service providers, URLs, port numbers,
protocols, path and names of the files to download).
Bandwidth Usage is also fairly required since the
main goal of FR is to download data from external
sites. Finally, Data Recovery and Error Notification
are also important since failed download activities
must be reported to the system administrator and to
the scheduler threads that may try to recover the lost
data.
Two main clusters of conflicts have been identified
for the system: (i) Bandwidth Usage versus Data
Format Interoperability / Data Delivery / Data
Recovery / Transfer Data Using a Web Service /
Transfer Data Using FTP Protocol and Transfer
Data Using HTTP Protocol: all download and data
delivery activities affect the network bandwidth
resource; (ii) CPU Usage versus Data Processing:
although minimum, Data Processing is required for
managing the scheduling of operations, thus
requiring the use of the CPU resource.
The following concerns have been identified as
crosscutting, as they interfere with multiple system
functionalities: Backup, Bandwidth Usage,
Configurability, Data Delivery, Data Format
Interoperability, Data Recovery, Error Notification,
Persistence, Scheduling, Transfer Data Using a Web
Service, Transfer Data Using FTP Protocol and
Transfer Data Using HTTP Protocol.
4.2.2 File Extractor and Transformer
Regarding the analysis of the File Extractor and
Transformer component, most significant negative
contributions are related with Bandwidth Usage and
CPU Usage. This pattern confirms the performed
analysis since the FET system is a major resource
user, thus affecting negatively these two concerns.
Five concerns have been identified as the most
required within the File Extractor and Transformer
system, namely: Configurability, Data Format
Interoperability, Data Recovery, Error Notification
and Validity.
These concerns are required very often since
they are strongly coupled to the Validity concern that
is an important issue for the FET system. The FET
system must, on the one hand, process all input files,
determining if the extracted data is valid or if the
input file format has changed. When detecting these
abnormal cases, the system must not be
compromised nor compromise other processing
requests (Data Recovery) and the system
administrator must be advised as soon as possible
that some input files are not being processed (Error
Notification). The FET component requires both
internal configuration (ETL script configuration for
each particular input file) and parameterization
(regarding the Data Delivery Web Service), which
justifies the strong need for the Configurability
concern. Finally, the FET component needs to
interchange data with data delivery interfaces and
with the FR component, so data communication
normalization is required (Data Format
Interoperability).
Regarding conflicts, these have been identified in
four match points: Computation in Parallel, Data
Delivery, Data Processing
and Transfer in Parallel.
From these the most representative clusters are: (i)
Bandwidth Usage versus Data Delivery / Data
Format Interoperability / Data Recovery / Transfer
in Parallel: Data Delivery usually refers to a
considerable amount of data requiring high
Bandwidth Usage. That is also affected by
interoperability constraints, possible recovery
actions and simultaneous transfer connection; (ii)
Data Recovery versus Response Time / Parallelism:
Data Recovery slows down Response Time while
Parallelism makes the system prone to error (due to
concurrent execution flows).
The following concerns have been identified as
crosscutting (as they affect multiple system
functionalities): Bandwidth Usage, Configurability,
Data Delivery, Data Format Interoperability, Data
Recovery, Error Notification, Transfer Data Using a
Web Service and Validity.
5 RELATED WORK
Some approaches have been proposed related with
the work presented in this paper. In (I. Brito, 2006)
an initial approach for the representation of a
Concern template is provided. However, this
approach is limited to the Concern concept (e.g. no
requirement information is available) and the
navigation between concepts is limited. On the other
ICEIS 2007 - International Conference on Enterprise Information Systems
78