I am looking at running the background job SDS tools sds_import and sds_migr on a large number of IS docs (> 100 million).
I was hoping to get an idea of how fast (docs/sec) each tool can process.
For sds_import, it would be running on a busy production system. Not sure if there is much speed difference between SDS storage types, but let's assume NetApp Snaplock here. Assume just one instance of sds_import, though if it scales well, that would be good to know. Based on limiting tested, it doesn't appear that sds_import uses page cache or the SDS workers. The IS activity seems to be solely updating doctaba and the mkf docs table.
For sds_migr, the source would be MSAR on a busy production system and the target would be a SDS library (let's assume Snaplock again) dedicated for migration (meaning no new docs are going to this library). Assuming 25 SDS Workers. Assume just one instance of sds_migr, though if it scales well, that would be good to know. My limited testing indicates that page cache does not get used but that some kind of direct copying occurs. Besides the copying, IS activity would presumably center on updating doctaba and mkf docs table.