Accelerating map cache creation by using file geodatabases

Discussion created by WillAllender on Feb 6, 2013
I am trying to accelerate map cache creation (and decrease Oracle/server load) by following esri's blog / best practices listed here:


Specifically, I am trying to use the recommendation to copy file gdbs to each server in the caching cluster.

I copied a file gdb to an identical location on each of 9 servers.  I have the "publish" version of the mxd on one of the 9 servers and I attempt to publish it.  I can successfully get through the analyze step, I have a validated data store, it succeeds in creating the staging file (according to the GP Results), but it fails on the upload service definition step, claiming that each of my 15 data layers in my mxd are "invalid".

I worked with esri tech support on the issue, but they were not really aware of this best practice, and struggled with the difference between "Publisher Folder Path" and "Server Folder Path" on the Data Store.  The only thing we could get to work was to share the folder that housed the file gdb, map the shared resource in ArcCatalog, and build the data store using the share name.  However, I don't think this is what the best practice intended. Instead of 9 servers pointing at an Oracle gdb, I now have 9 servers pointed at a single file gdb on one of the servers.  Not ideal.

Has anyone implemented this "best practice" and/or is there more detailed documentation on the steps?  Esri Tech Support is not aware of any and claim that it is only a blog, so they don't have any detailed information on how to support it.