/[volute]/trunk/projects/registry/VODataService/VODataService.tex
ViewVC logotype

Diff of /trunk/projects/registry/VODataService/VODataService.tex

Parent Directory Parent Directory | Revision Log Revision Log | View Patch Patch

revision 5164 by msdemlei, Fri Oct 5 11:27:11 2018 UTC revision 5165 by msdemlei, Mon Oct 8 13:53:51 2018 UTC
# Line 2  Line 2 
2  \input tthdefs  \input tthdefs
3  % widen up the display a bit so that 75 column listings still fit on  % widen up the display a bit so that 75 column listings still fit on
4  % the page  % the page
5  \usepackage[width=14cm]{geometry}  \usepackage[width=14cm,left=4cm]{geometry}
6  \usepackage{listings}  \usepackage{listings}
7  \lstloadlanguages{XML}  \lstloadlanguages{XML}
8  \lstset{flexiblecolumns=true,tagstyle=\ttfamily,showstringspaces=False}  \lstset{flexiblecolumns=true,tagstyle=\ttfamily,showstringspaces=False}
# Line 82  Line 82 
82  eInfrastructure Science Repositories initiative.  eInfrastructure Science Repositories initiative.
83    
84  Version 1.2 of this document was developed in part with support from the  Version 1.2 of this document was developed in part with support from the
85  German federal ministry for research and education's e-inf-astro (BMBF  German federal ministry for research and education's e-inf-astro project (BMBF
86  FKZ 05A17VH2)  FKZ 05A17VH2).
87    
88    
89  \section*{Conformance-related definitions}  \section*{Conformance-related definitions}
# Line 207  Line 207 
207  \citep{std:XML}, XML Schema \citep{std:XSD},  \citep{std:XML}, XML Schema \citep{std:XSD},
208  and VOResource itself.    and VOResource itself.  
209    
210    \subsection{Additional Use Cases for Version 1.2}
211    
212    In the following, we collect use cases that guided the development of
213    VODataService to its version 1.2.  We do not formally derive
214    requirements from them but briefly note which new features enable or
215    facilitate the specific use case.
216    
217    A few of the changes are necessary for consistency with other standards
218    such as TAP (extendedType interpretation, requirement to use ADQL
219    delimited identifier literals in names where appropriate) or VOTable
220    (arraysize interpretation).  These were obviously not guided by specific
221    use cases.
222    
223    
224    \paragraph{What services have data for the Crab nebula covering the H$\alpha$
225    line taken in the second half of 2015?}  In version 1.1, this use case
226    would have been covered by the \xmlel{stc:STCResourceProfile} type,
227    which was never properly standardised or widely adopted.  In the current
228    version, the \xmlel{spatial}, \xmlel{spectral}, and \xmlel{temporal}
229    children of \xmlel{coverage} enable discovery by coverage on the various
230    axes.  It is worth noting that the spectral coverage is for the solar
231    system barycenter, so this use case does \emph{not} immediately enable
232    the discovery of, say, H$\alpha$ images of remote galaxies.  Redshift
233    correction has to be applied by the client based on knowledge about the
234    object(s) investigated.\todo{Also, we don't do solar system well
235    spatially.  Mention this here?  Fix it now?}
236    
237    \paragraph{Find all ObsCore services publishing data taken at the
238    Telescope X.} This use case could be satisfied in version 1.1 through
239    the use of \xmlel{vs:DataCollection} records and relationships to the
240    respective TAP services.  However, this scheme led to error-prone query
241    patterns, and few such data collections were actually registered; see
242    the IVOA Note on Discovering Data Collections \citep{note:DataCollect} for
243    details.  To better support the scheme proposed there, version 1.2 adds
244    the \xmlel{vs:TODOTODO} type that identifies a resource as data-like but
245    allows the addition of various capabilities to the record (which
246    \xmlel{vs:DataCollection} did not).  An analogous use case would be
247    ``Find all TAP services publishing tables from Gaia DR2''.\todo{I need a
248    name for these resource types.  Desperately.}
249    
250    \paragraph{Find a large-scale survey of sources between 20 and 40 GHz.}
251    While the spectral constraint is easily satisfied by the new coverage
252    children, the ``large-scale'' part is much harder to operationalise.
253    However, the plain table size often is a useful proxy in such discovery
254    problems.  The new \xmlel{nrows} child of \xmlel{vs:Table} communicates
255    it.
256    
257    \paragraph{Find a resource that has sources in M51 down to 27 mag in V.}
258    The constraint about finding a resource that has V magnitudes for M51 is
259    expressible using spatial coverage and the column's UCDs.  To express
260    something like ``down to $27^{\rm m}$'' one would at least need
261    VOTable-style \xmlel{VALUES} children for columns.\todo{Do we want
262    that?}
263    
264    \paragraph{Plan a cross-service query.} Systems like OGSA-DAI
265    \citep{2011ASPC..442..579H} perform orchestration of SQL-like queries
266    between multiple services automatically, in particular cross-service
267    JOINs.  In order to work efficiently, such services need column
268    statistics like histograms and the percentage of NULL values.\todo{Do we
269    want this?  If so, what exactly?}
270    
271    \paragraph{Find services serving time series.} In the current registry
272    model, users looking for spectra would select SSAP services.  With the
273    growing adoption of ObsCore (and a growing number of services abusing
274    SSAP to serve time series), the model of selecting data types by
275    constraining the service protocol no longer works; in the ObsCore
276    example, clients now have to query all services and constrain the
277    \verb|dataproduct_type| column. However, for dataset types not overly
278    common, well more than 90\% of the services could be excluded without
279    sending a query there based on a declaration of dataset types available
280    in the Registry.\todo{Actually, a VOTable VALUES workalike would be
281    enough for that and ObsCore.  But: SIAv2?  Data collections, perhaps?}
282    
283    \paragraph{Facilitate discovery of full DALI services.}  The issue here
284    is that DALI forsees synchronous and asynchronous endpoints as the
285    standard case for many protocols -- it already is standard for TAP.  The
286    current way to discover these, together with the pertinent VOSI
287    endpoints, is built on \xmlel{vs:ParamHTTP}-typed interfaces and has
288    known weaknesses, in particular when multiple authentication schemes
289    necessitate multiple sync/async pairs.\todo{See what UWSRegExt brings.}
290    
291    
292    
293  \section{The VODataService Data Model}  \section{The VODataService Data Model}
294    
# Line 334  Line 416 
416  collections of data, perhaps available through ftp services or offline.  collections of data, perhaps available through ftp services or offline.
417  It might still be used as such, although the utility of such a  It might still be used as such, although the utility of such a
418  registration is questioinable at present.  With the Endorsed Note on  registration is questioinable at present.  With the Endorsed Note on
419  discovering data collections \citep{todo:DDC}, a more capabable type for  discovering data collections \citep{note:DataCollect}, a more capabable type for
420  reigstering data collections independent of services is necessary, and  reigstering data collections independent of services is necessary, and
421  in general it is desirable to attach actionable properties (i.e.,  in general it is desirable to attach actionable properties (i.e.,
422  capabilities) to resource records.  Therefore, version 1.2 of  capabilities) to resource records.  Therefore, version 1.2 of
# Line 1174  Line 1256 
1256    
1257    
1258  The \xmlel{vs:TableSchema} type collects  The \xmlel{vs:TableSchema} type collects
1259  tables together that are logically related.  For example, a single  tables together that are logically related.\todo{Better cater for the
1260    case that tables are outside of all schemas (see SIMBAD)}  For example, a single
1261  resource may provide access several major astronomical catalogs  resource may provide access several major astronomical catalogs
1262  (e.g. SDSS, 2MASS, and FIRST) from one site, enabling high-performance  (e.g. SDSS, 2MASS, and FIRST) from one site, enabling high-performance
1263  cross-correlations between them.  Each catalog can be described in a  cross-correlations between them.  Each catalog can be described in a
# Line 1803  Line 1886 
1886  use of \verb|xsi:type="vs:ParamHTTP"| indicates that the interface  use of \verb|xsi:type="vs:ParamHTTP"| indicates that the interface
1887  accessed via the URL given by the \xmlel{accessURL}  accessed via the URL given by the \xmlel{accessURL}
1888  element complies with the general parameter-based protocol described  element complies with the general parameter-based protocol described
1889  in this section.    in this section.\todo{resultType stinks when we have DALI
1890    RESPONSEFORMAT.  Would anyone miss it?  If so, how can we fix it?}
1891    
1892    
1893    
# Line 2724  Line 2808 
2808         \xmlel{vs:VOTableType}.         \xmlel{vs:VOTableType}.
2809  \end{itemize}  \end{itemize}
2810    
2811  \bibliography{ivoatex/ivoabib,ivoatex/docrepo}  \bibliography{ivoatex/ivoabib,ivoatex/docrepo,local}
2812    
2813  \end{document}  \end{document}

Legend:
Removed from v.5164  
changed lines
  Added in v.5165

msdemlei@ari.uni-heidelberg.de
ViewVC Help
Powered by ViewVC 1.1.26