I have an application for geoserver that will include accessing hundreds of different workspaces that are each based on an OracleNG schema. One problem I have is that it takes forever for geoserver to start with a large number of workspaces because of the database activity initializing each workspace. I’m hoping some kind of lazy initialization might be possible. But the other problem I have is that with hundreds of workspaces and thousands of users I need to plan for scalability. That might include deploying multiple geoserver instances in a load balanced environment.
With each geoserver keeping workspaces in the file system, only one specific geoserver instance can access a given workspace instance. How is this usually setup? Do I need to dedicate each instance to specific workspaces and turn off global workspaces so Apache can examine the incoming URL and map it to the right instance?
My application includes a service that creates new workspaces and layers. If I can’t use global workspaces and need to map to a specific geoserver instance then the code that sets up workspaces will also need to get involved in somehow telling apache to map the URL also. Gets messy and I’m looking for ideas.
Thanks for any comments.
Walter Stovall
Byers Engineering Company