1 Meeting on the Management of Statistical Information Systems (MSIS 2010) SDMX architecture for data sharing and interoperability Francesco Rizzo, ISTAT, Italy Adam Wronski, Eurostat Daejeon, Republic of Korea, 26-29 April 2010
Jan 17, 2016
1
Meeting on the Management of Statistical Information Systems (MSIS 2010)
SDMX architecture for data sharing and interoperability
Francesco Rizzo, ISTAT, ItalyAdam Wronski, Eurostat
Daejeon, Republic of Korea, 26-29 April 2010
2
SDMX service infrastructure: rational and keywords
The idea is based on the analysis of the current level of
SDMX implementations within Member States and their
needs expressed during several TF, WG and other
meetings on the possibility to re-use software already
developed with the main aims of reducing costs and
increasing productivity
3
SDMX can help by ….
improving quality and efficiencies in the exchange and
dissemination of data and metadata;
reducing national reporting burden;
reducing costs through the re-use of the software;
facilitating and standardizing the use of new
technologies as XML and Web services.
4
Data Repository (Warehousing) Architecture
NSI
EurostatPull Requestor
eDAMIS
Data reception
SDMX Registry
Loadingpreparation
Verification /Conversion
o SDMX
Receiveddata in
SDMX-MLLoader
register
DataWarehouse
Database
query
Dissemination
XSL forSDMX-ML
PULL
PUSH
5
Data Hub Architecture
Dissemination
XSL forSDMX-ML
Data Hub
QuerySDMX
messages
cache
WebService
WebService
WebService
GUI
Data Providing Organizations Data collector Organization
6
Eurostat strategies to foster SDMX implementations within NSIs
A project aiming to design an SDMX service infrastructure
for NSIs and developing related building blocks;
Finance and support, through SDMX ESSnet , a group of Member States that have joined their resources in order to develop SDMX re-usable software.
Capacity building actions
7
SDMX NSI reference service infrastructure:main objectives
support Census Hub and other Eurostat project
facilitate SDMX implementation within NSIs with a particular attention to PC-AXIS community
stimulate a “SDMX community of developer”
8
SDMX NSI reference service infrastructure:deliverables
SDMX NSI reference service architecture document It represents the syntheses of several experiences worldwide and may be considered not as a strict specification but rather a guide or “best practice” document; It provides a description/specification of a generalized infrastructure that could be re-used partially or entirely by NSIs interested in SDMX projects;
A set of software building blocks;
Mapping Assistant tool;
Capacity building actions (Training and Technical Workshops).
9
SDMX NSI reference service infrastructure
Hub
PullRequestor
Data Collector SDMX NSI InfrastructureNSI
Disseminationenvironment
MappingStore
MappingAssistant
NSI Web Service
Web ServiceProvider
DataRetriever
SDMX DataGenerator
SDMXQuery Parser
DDB
Pull
DSDs
10
The mapping process
Data within Data Providers’ dissemination databases are described using “local” structural metadata (concepts, code lists, formats);
“local” structural metadata and SDMX structural metadata must be mapped:
concepts mapping
codes mapping
11
Concept and code mapping: an exampleLOCAL CONCEPTS
LOCAL CODES
SDMX CODES
SDMX CONCEPTS
Freq 12 M FREQ
IT REF_AREA
Type m W ADJUSTMENT
DomainSetCategory
eip63
PROD STS_INDICATOR
Ateco DADB
N100DAN100DB
STS_ACTIVITY
1 STS_INSTITUTION
Um pe 2000 STS_BASE_YEAR
PURE_NUMB UMIS
0 UMIS_MULT
Year, Month 2005, 3 2005-03 TIME_PERIOD
P1M TIME_FORMAT
Value OBS_VALUE
A OBS_STATUS
12
The Mapping Assistant tool: workflow
Map localDatabase schema
(Dataset)
Map localConceptsTo DSD
(Mapping Set)
Map localCodes
To DSD(Transcoding)
13
The Mapping Assistant tool: Dataset
14
The Mapping Assistant tool: Mapping Set
15
The Mapping Assistant tool: Transcoding
16
Software maintenance and governance
The software is offered as open source package under the EUPL licence
Up to now Eurostat is managing both the evolutive and adaptive maintenance.
How the versioning will be handled when NSIs will re-use and improve the software?
Is the model used by many “open source” community applicable to the “statistical community”?
know-how could come from NSIs participating in two ESSnet projects