I have been looking for some statistics on differences in efficiency for
Geoserver in providing map tiles or arbitrary maps. Tile caching is not an
option in one of our projects due to the use of SQL parameter WMS services
(http://www.mindland.com/wp/?p=121).
Any statistics or sources for
If you truly can't do any caching then I'd guess arbitrary maps would
perform slightly better. Less requests to the server for each view
you're looking at.
But you probably could cache tiles in your use case. Could use
parameter filters in GeoWebCache -