[mapserver-users] SUMM: CGI Map changes and question

[Lowell.Filak]

What if your master shapefile(s) where created as a tile index that
pointed to the individual farm shapefiles. If you included an item in the
tile index for the farm number, you could set the index as a normal
polygon layer and query against that item to quickly set the extents for a
particular farm and then also have a layer that uses the index as a true
tile index which would then point to the shapefiles of the individual
farms. When doing a spatial query the tile index will limit the query to
only the individual farm (roughly depending on shape) and an attribute
query could be forced to query only the shapefile for a particular farm
by setting the data for the layer to the individual shapefile that holds
data for that farm. That should be less messy and still increase the
speed.
Lowell F.

On Tue, 9 Apr 2002, Robert Crossley wrote:

>
> Thanks for your replies.  This is an interesting experience participating in a list as a complete newbie.
>
> Form the replies, I understand that Mapserver doesn't maintain knowledge of each users current state, so it means that any changes to a map template need to be made for
> each call.  No great problem.  Alternatively I could use the SaveMap call to save the current settings in a temporary Map file for use in that user-session.  At one point I thought I
> would have to generate the Map file dynamically prior to calling MapServer, so both of these options are much slicker.
>
> I thought that other web servers like MapExtreme and ARCIMS spawned new instances for a user/ session and built on previous changes, but thus had the potential to bog
> down under load.  MapServer's stateless approach avoids this load, but does it have the disadvantage that it needs to redo operations like filters each time?
>
> My app has about 1500 farmers who each want to look at only their farm.  There are 3 master tables of farm data: paddocks, harvested areas and labels, each with about ~
> 30,000 records.  Each record in these tables is tagged with the farm number.
>
> I have been able to change mapfile data parameters using a URL OK, so I can allow each farmer to look at only their data by pre-processing the master table and splitting it into
> separate farm tables that have the farm number as part of the file name, and changing the data reference for the farm layer for each farmer at run-time.
>
> Another option is to filter the master table in MapServer and change the filter string based on farm number.
>
> While the first option is messy terms of the number of files, it is fast.  The filter option looks neater, but I am cautious that it will load the server substantially when a large number
> of farmers are using it as I gather that the filter has to be applied in each call.
>
> Does anyone have any comments on the performance of filters  and the load it places on the server?  The 3 master shape files are about 9MB each versus  about < 10KB each
> for the shape files for each farm.
>
> Thanks
>
> R
>
>
> >
>
> Robert Crossley
> Robert Crossley & Associates
> 9 Short St
> New Brighton NSW 2483
> AUSTRALIA
>
> P: 02 6680 1309
> F: New Connection
> M: 0419 718 642
> E: robert at wotzhere.com
>
>
>