Large Map Cache Generation - Resuming a failed generation vs. On Demand caching?

Discussion created by dbowerman on May 9, 2013
Latest reply on May 10, 2013 by JQuinn-esristaff
Hello ESRI Community,

For several weeks I have been struggling to generate a fairly sizable map cache. The anticipated size of the cache will be about 800GB, with the final level, 14, estimated to be about 530GB.

This level 14 generation has failed on me a few times, and this typically results in an ArcCatalog crash. Which means I do not see errors in ArcCatalog about why it crashed. It's taken several days to generated 365GB of the 530GB level.

I ran again with "RECREATE_ALL_TILES", and several days later, it crashes again without a good error.

Seemingly running the cache creation with "RECREATE_EMPTY_TILES" -- this similarly takes an extremely long time while seemingly not growing the cache size at all.

I only seem to get farther along if I delete Level 14 altogether and start over. Because this takes several days per trial, I don't have a lot of attempts under my belt. I wanted to ask here before I continued with this. Two questions:

1. Is there any true way to "resume" a crashed map cache without seemingly starting over? This looks like a good solution if I were receiving errors for certain extents: http://resources.arcgis.com/gallery/file/geoprocessing/details?entryID=5E1F2FE3-1422-2418-88F7-285291B8031D

2. We have had tremendous lag issues with an On Demand map cache in our previous experience, but I would like to look at this again. Since we have a significant portion of the cache generated already, is it possible to use this partial cache and let time "fill in the blanks" by allowing users to generate new tiles on demand?

Your insights and expertise are much appreciated, thank you in advance.