[mapserver-users] Raster performance hints

Travis Kirstine traviskirstine at gmail.com
Tue Nov 6 12:05:46 PST 2018


What you describe is pretty much the approach we take.  If you are using
geotiffs ensure that they are internally tiles (-co TILED=YES) and the you
are creating a shptree index on the tile index as noted in the docs.  At
some point you may want to consider creating a lower resolution mosaics of
coverages and using GROUPS and MIN/MAXSCALEDENOM to load the layers at
different scales.  Make sure to declare the layer's extent as well

Something like:

    LAYER
        NAME "full_res_10cm_tiles_withovr"
        EXTENT 607500.000000 4824000.000000 654000.000000 4859500.000000
        TYPE RASTER
        PROCESSING "RESAMPLE=AVERAGE"
        TILEINDEX /maps/tindex.shp
        TILEITEM location
        GROUP "my_photo"
        MAXSCALEDENOM 10000
        OFFSITE 0 0 0
        METADATA
            "wms_extent" "607500.000000 4824000.000000 654000.000000
4859500.000000"
        END
        PROJECTION
            "init=epsg:2958"
        END
    END
    LAYER
        NAME "resampled_2m_tiled_bigtif_withovr"
        EXTENT 607500.000000 4824000.000000 654000.000000 4859500.000000
        TYPE RASTER
        PROCESSING "RESAMPLE=AVERAGE"
        DATA /maps/2m_overall.tif
        GROUP "my_photo"
        MINSCALEDENOM 10000
        MAXSCALEDENOM 65000
        OFFSITE 0 0 0
        METADATA
            "wms_extent" "607500.000000 4824000.000000 654000.000000
4859500.000000"
        END
        PROJECTION
               "init=epsg:2958"
        END
    END





On Tue, 6 Nov 2018 at 13:39, Ziegler Stefan <Stefan.Ziegler at bd.so.ch> wrote:

> Hi
>
> We are using MapServer for vector data stored in a PostgreSQL/PostGIS
> database. For a new project we are planning to provide a WMS based on
> raster layers and therefore in need of some performance hints or your
> experiences.
>
> The raw data will be a lot of single GeoTIFF (aerials, hillshading,
> slopes, etc.) files, like 5'000 - 10'000 from different organizations.
> Since the originate from different organizations it's possible that they
> can have different resolutions. Conceptionally for aerials it would look
> like this:
>
> Aerial group
> |_ aerial from organization one (1000 single tiffs, resolution 10cm,
> approx 10k x 10k pixels)
> |_ aerial from organization two (600 single tiffs, resolution, 12.5 cm,
> approx 10k x 10k pixels)
> |_ aerial from organization three (3000 single tiffs, resolution, 10 cm,
> approx 10k x 10k pixels)
>
> For some hundred single GeoTIFFs we have good experience with the
> Tileindex approach and an handmade external overview that kicks in at
> smaller scale.
>
> Will this approach still work with a massive amount of single tiffs? Or
> are the better ones? What about the compression (quality vs. decompression
> speed vs. disk space?) Or would you expect a much better performance with -
> let's say - a BigTIFF for each organisation?
>
> Thanks for any hints.
>
> Best regards
> Stefan
>
> Freundliche Grüsse
>
> Stefan Ziegler
> Kantonsgeometer / Leiter Amt für Geoinformation
>
> Amt für Geoinformation
> Rötistrasse 4
> 4500 Solothurn
>
> Telefon +41 32 627 75 96
> stefan.ziegler at bd.so.ch
> http://www.so.ch
>
>
> _______________________________________________
> mapserver-users mailing list
> mapserver-users at lists.osgeo.org
> https://lists.osgeo.org/mailman/listinfo/mapserver-users
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/mapserver-users/attachments/20181106/239e262e/attachment.html>


More information about the mapserver-users mailing list