[Mapserver-users] Fastest Data Format

[GAIA]

>ArcIMS seems to handle SDE just fine, but that should be no surprise :) 
>I imagine that ArcIMS uses connection pooling to avoid the overhead 
>associated with making SDE connections.
>
>I should reiterate at this point that the performance problem does not 
>adhere to SDE necessarily or to Mapserver necessarily, but to the amount 
>of time it takes to negotiate a connection with an SDE server. Once the 
>connection is made, everything is performant. It is making the 
>connection which is the issue. If you are only making one connection, 
>probably it is completely reasonable.
>
>However, the current design of Mapserver treats each layer as a distinct 
>entity, and each layer datasource as a distinct entity. So for each 
>SDE-based layer, there is a separate SDE connection to be made. Over 
>numerous layers, things can add up to quite a long delay.
>

As discussed before, you are absolutely right.  How many of us create mapfiles with only one layer?  None of course.  A GeoSpatial store typically houses vast quantities of data and layers.  Pragmatically, SDE is unusable through a MapServer interface at this time.  So it does come back to performance--through connection volumes as you indicate, but nonetheless collective performance is throttled.  



>Jan Hartmann has made a proposal for connection pooling between layers, 
>I do not know if he has a preliminary implementation or if such an 
>implementation will be included in 3.7.
>
>Paul

I have always been a proponent of MapServer and fended off the ESRI sales force trying to sell me ArcIMS.  We still have a stonger more extensible product in MapServer thanks to all the dev. team.  Let's look at filling this hole. 

Has there been any progress on this issue?  We would be more than happy to help in any way we can.


Cheers,
GAIA