We have a staging cached map service and an identical production cached map service on the same instance, but they reside in two different server folders. Our workflow consists of building the cache update in the staging service and then importing using geoprocessing tools into the production map service. Occasionally, this results in errors and the ability to troubleshoot errors is limited. It is unknown whether errors occur because of minute inconsistencies between data frame extents for published map documents or in the image format type. When reviewing an esri user conference presentation I came across this slide and was curious if anyone knows or is familiar with this workflow of having two different services with a shared cache folder. If you know about this, how has it worked for you? Do you have a better scenario?