Hmm. That seems strange given that (at 10.1) default GEOMETRY_STORAGE is GEOMETRY.
Anyway, I was able to get my featureclasses to use SDEBINARY simply by
(1) Changing the DEFAULTS GEOMETRY_STORAGE config parameter to SDEBINARY
(2) Re-running my Python script that creates the feature class(es) from our 3rd party data providers [In my case, I am lucky because my workflow includes this step to RECREATE all my feature classes].
Note: If what you are saying is true (Not sure why our ESRI technical guy did not alert me when I told him I was planning to run MIGRATE STORAGE to go from GEOMETRY to SDEBINARY) that migrate storage does not support migrating TO SDEBINARY,
then folks should (a) Export data to some temporary file geodatabase, (b) Change DEFAULTS GEOMETRY_STORAGTE config param to SDEBINARY, then (c) copy all your data back to your geodatabase from the temporary file geodatabase.
Just also thought I should SHARE with this community that I have found performance to improve by about 100% when I moved to SDEBINARY. My performance test was on a point featureclass of > 650,000 points drawn at full extent using the
arcgis share as (wizard) to publish a map service, using the PREVIEW tool to indicate how many seconds it takes to load the map
with this single feature class.
Perhaps some ESRI geodatabase person can reply and tell us why this is the case.
Curious why ESRI has decided to make default GEOMETRY_STORAGE = GEOMETRY for SQL SERVER when SDEBINARY is faster.
Thanks,
Anastasia