Sorry yuriesky for the unintended hijack of your thread... Both arguments are very interesting... and I am interested in both of them...
Probably we could split the thread. Guys, do you follow any convention in this list for splitting treads ? On Mon, Jul 19, 2010 at 11:18 PM, salas <fsa...@geocuba.cu> wrote: > Hi, this topic is very interesting but in my situation we need to know > about > the vectorial data, not storing of raster data. > could someone tell me about this? store limits for vectorial data. > > yuriesky > > ----- Original Message ----- From: "Pierre Racine" < > pierre.rac...@sbf.ulaval.ca> > To: "PostGIS Users Discussion" <postgis-users@postgis.refractions.net> > Sent: Monday, July 19, 2010 10:25 AM > Subject: [!! SPAM] Re: [postgis-users] may have postgis store limits for a > > big project? > > > >We do not know yet.. it could be even hundreds ! >> >> WKT Raster is not really limited in size. But we still have to test on >> datasets the size you are using. >> >> each region (european country, US state, province, etc...) would have few >>> raster (even 20) >representing different kind of information... (each >>> pixel would be 5mx5rm)... >>> >> >> With WKT Raster, you can store all those rasters as a unique tiled >> coverage. In Oracle probably you stored them one table per raster? >> >> and I need to know the value of a pixel for all the rasters in that >>> particular coordinates... >>> >> >> For this you typically use ST_Value() like this: >> >> SELECT ST_Value(rast, ST_Geomfromtext('Point(-78.1 58.1)', 4326)) >> FROM srtm_tiled_100x100 >> WHERE ST_Intersects(rast::geometry, ST_Geomfromtext('Point(-78.1 58.1)', >> 4326)) >> >> One query like this takes 16 milliseconds on my machine on a 100 pixels x >> 100 pixels tiled coverage. There are 46800 tiles. >> >> Don't forget to assign a reference system to your raster when converting >> it to .sql (-s gdal2wktraster's option) and to construct an index on the >> table (-I option). >> >> with Oracle 11g GEORaster it is possible to compress the rasters (and get >>> lower performance of >course)... in my case the compression rate is up to >>> 600 !!!... so the amount of data is >drastically reduced... >>> >> >> WKT Raster relies on PostgreSQL for compression. I don't know the >> compression rate. >> >> nevertheless I need to be able to do a few (up to 20) queries like this >>> at >>> the same time in in a >fraction of second...: select pixel from raster >>> where (x=X and y=Y). One query for each raster, >representing a different >>> information >>> >> >> If you tile and index everything properly that should not be a big deal. >> Let us know! >> >> If WKTRaster is mature enough, we can create a cluster without the need >>> of >>> paying a prohibitive license per CPU as for Oracle ! >>> >> >> If you save extra money by not having buying an Oracle license and you are >> interested in collaborating in the development of WKT Raster, let us know >> ;-) >> >> Pierre >> >> Sebastian >> On Mon, Jul 19, 2010 at 5:00 PM, Pierre Racine >> <pierre.rac...@sbf.ulaval.ca> wrote: >> Hi Sebastian, >> >> What kind of query are you planning on the raster? You know you can also >> use WKT Raster to simply "register" filesystem rasters, creating a king of >> geospatial catalog, without actually storing them in the database... >> >> How many 16GB raster do you have? Are your raster all representing the >> same theme? Are they overlapping (or they form a regular grid)? >> >> We don't know of that many users cases up to now, but WKT Raster prooved >> to be very stable working on 1 GB raster coverage (see the tutorial). >> >> Pierre >> >> From: postgis-users-boun...@postgis.refractions.net >> [mailto:postgis-users-boun...@postgis.refractions.net] On Behalf Of >> Sebastian E. Ovide >> Sent: 19 juillet 2010 11:34 >> To: PostGIS Users Discussion >> Subject: Re: [postgis-users] may have postgis store limits for a big >> project? >> >> Hi All, >> >> I am interesting too... I am considering PostGIS+WKTRaster instead of >> Oracle GeoRaster... but for that can happen, Postgresql must be able of >> managing some simple tables of 30+M (Millions) rows plus a lot of Rasters >> of 16GB each... (16GB not compress data... so are huge images !). Each >> query (simple look up) must response in a fraction of second.... >> >> any successful case ? >> >> Thanks ! >> On Fri, Jul 16, 2010 at 1:13 AM, salas <fsa...@geocuba.cu> wrote: >> Hello to all: >> I am working in a project where they are managed more than 20 geoespatials >> thematics. Each thematic one has a considerable volume of since >> information it is of the whole country (I am speaking of a lot more than a >> million of records). In the project we need to make (mostly) intersection >> consultations keeping in mind literal attributes. >> I need to know the experience of somebody in a project of this span and if >> PostGIS would present some limitation therewith. >> >> regards yuriesky >> >> _______________________________________________ >> postgis-users mailing list >> postgis-users@postgis.refractions.net >> http://postgis.refractions.net/mailman/listinfo/postgis-users >> >> >> >> -- >> Sebastian E. Ovide >> >> >> >> _______________________________________________ >> postgis-users mailing list >> postgis-users@postgis.refractions.net >> http://postgis.refractions.net/mailman/listinfo/postgis-users >> >> >> >> -- >> Sebastian E. Ovide >> >> >> >> _______________________________________________ >> postgis-users mailing list >> postgis-users@postgis.refractions.net >> http://postgis.refractions.net/mailman/listinfo/postgis-users >> >> > > _______________________________________________ > postgis-users mailing list > postgis-users@postgis.refractions.net > http://postgis.refractions.net/mailman/listinfo/postgis-users > -- Sebastian E. Ovide
_______________________________________________ postgis-users mailing list postgis-users@postgis.refractions.net http://postgis.refractions.net/mailman/listinfo/postgis-users