[OSGeo-Boston] Switching From ArcIMS to "something else"

[Saul Farber]

Actually, MassGIS has done a pretty good job of using FOSS in a 'hybrid'
way and the use of geoserver with ArcSDE is fairly well supported.

My take on how to 'web-enable' an ArcSDE based storage would be
something like this:

1)  Install GeoServer
2)  Install the GeoServer ArcSDE support
3)  Pick which layers you want to serve and create 'featuretypes' for
them in GeoServer
4)  Create styles for each featuretype (perhaps using udig's SLD editor)
5)  Apply the styles to the featuretypes
6)  Go use the built-in openlayers 'demo client' to see your data.
7) (optional) hit 'save source...' on your demo client page and start
hacking!


--saul


On Wed, 2007-11-21 at 10:01 -0500, Christopher Schmidt wrote:
> So, I got 3 personal (count 'em, 3) emails yesterday on what to switch
> away from an ArcIMS server solution to. Since I hate answering anything
> in private (shared knowledge > non-shared knowledge), I thought I'd
> address them here, since I know that a fair number of the local people
> work in/around MIT, which uses (as I understand it) ArcIMS to do their
> WMS-style work.
> 
> The first question to answer when switching away from an ESRI-based
> solution is what type of data you're serving from. Moving away from ESRI
> can be hard if you've got all your data in a giant ArcSDE database --
> not impossible, but it makes things more difficult. If, on the other
> hand, you have things like shapefiles, then it may make the switch
> easier.
> 
> Next, figure out what you want to do with the data: Do you want to serve
> it up to GIS clients (ArcGIS, etc.)? Do you want to create a web
> interface? Do you want to help people to create mashups? Do you want to
> let users edit it? etc. 
> 
> If I were to take something like MIT's 'whereis' map datasource, I would
> probably (personally) do something like:
>  * Dump the data to a PostGIS interface
>  * Load it up into MapServer (this requires converting the existing
>    style information into MapServer's mapfiles, which is possibly
>    time-consuming)
>  * Set up a WMS against that, using MapServer
>     * This will be used by clients who consume WMS, a la ArcGIS/qgis
>  * Set up TileCache in front of it, picking a small number of
>    projections that are acceptable to pre-cache. I'd probably just build
>    this on a machine with lots of disk: it seems unlikely that you'd get
>    more than 100GB of cache even with three different layers. These
>    TileCache layers are 'basemaps' -- you probably want EPSG:4326 (for
>    most 'unprojected' maps), EPSG:900913/SphericalMercator (for use with
>    Google Maps), and the local projection (for local-only maps).
>  * Use OpenLayers to load the relevant layers into a map
>  * Use either a small overlay WMS layer or OpenLayers vector support to
>    display selected features and the like.
>  * Stand up FeatureServer, with attribute querying on useful attributes,
>    to get WFS-like services for GeoJSON, GeoRSS-Atom, KML, etc. This can
>    also have read-write layers for user input. 
>  
> Then, you document all these things and see who comes along and uses
> them :)    
> 
> Specific use cases, of course, matter. If your primary goal is to make
> data available for use in ArcGIS, this might not work for you. But
> MapServer (5.0) + TileCache + OpenLayers makes a great, fast map
> browser: see http://boston.freemap.in/ for an example of my personal
> attempts at that.
> 
> Regards,