218
SYLVIE POULIQUEN
There are three main characteristics for a data management system:
1. Its architecture
2. The quality control procedures
3. Data format and metadata attached to the data
5.1
System architectures
A data management system is designed according to the type of data
handled (images/profiles/timeseries/kilobytes versus gigabytes, etc), the
users access needs (individual measurements, geographical assess, integrated
datasets, etc), the level of integration needed, etc.
In the past decade, with the improvement of the computer technology,
the internet revolution, the increase of network speed and capacity, data
management systems have been progressively moving from centralized to
distributed systems. Two main architectures are nowadays commonly used:
¾ Distributed processing and centralized distribution: data are
processed in different places and are than copied in a single place for
distribution to users.
¾ Distributed processing and distribution: data are processed in
different paces and stay where they are. To ease user access a virtual
WWW portal is implemented that use networking techniques to find
the data that fit the user needs.
Each system has its advantages and drawbacks, depending on the type of
datasets to distribute and the contributors to the network. These different
architectures will now be quickly described through examples operating at
present.
5.1.1
ARGO data system: Distributed processing and centralized
distribution
Within the ARGO data system, the float data processing is distributed
among the contributing national data centres. They feed two global data
centres (GDACs) automatically with the latest version of their float profiles,
trajectories and metadata. Both GDACs are updated simultaneously to
ensure consistency between the two datasets. They synchronise their
holdings each night [in case a DAC (Data Assembly Centre) has updated one
GDAC and not the other one].
Individual agencies (some acting on behalf of several countries) assemble
the data collected from the communications system and carry out the initial
processing of the data. Each file is under the responsibility of a single DAC
(i.e. the data provider) who guarantees the quality and integrity of the data.
SYLVIE POULIQUEN
There are three main characteristics for a data management system:
1. Its architecture
2. The quality control procedures
3. Data format and metadata attached to the data
5.1
System architectures
A data management system is designed according to the type of data
handled (images/profiles/timeseries/kilobytes versus gigabytes, etc), the
users access needs (individual measurements, geographical assess, integrated
datasets, etc), the level of integration needed, etc.
In the past decade, with the improvement of the computer technology,
the internet revolution, the increase of network speed and capacity, data
management systems have been progressively moving from centralized to
distributed systems. Two main architectures are nowadays commonly used:
¾ Distributed processing and centralized distribution: data are
processed in different places and are than copied in a single place for
distribution to users.
¾ Distributed processing and distribution: data are processed in
different paces and stay where they are. To ease user access a virtual
WWW portal is implemented that use networking techniques to find
the data that fit the user needs.
Each system has its advantages and drawbacks, depending on the type of
datasets to distribute and the contributors to the network. These different
architectures will now be quickly described through examples operating at
present.
5.1.1
ARGO data system: Distributed processing and centralized
distribution
Within the ARGO data system, the float data processing is distributed
among the contributing national data centres. They feed two global data
centres (GDACs) automatically with the latest version of their float profiles,
trajectories and metadata. Both GDACs are updated simultaneously to
ensure consistency between the two datasets. They synchronise their
holdings each night [in case a DAC (Data Assembly Centre) has updated one
GDAC and not the other one].
Individual agencies (some acting on behalf of several countries) assemble
the data collected from the communications system and carry out the initial
processing of the data. Each file is under the responsibility of a single DAC
(i.e. the data provider) who guarantees the quality and integrity of the data.