Downloading and restoring simultaneously is the practical move — replaying from block 1 on a 1TB drive would take forever with the current chain size. I've been watching the api.steemit.com internal errors lately and it's clear the infrastructure load is increasing. Are you using LZ4 or ZSTD compression on the block data during the restore phase? Also curious whether the restored node will eventually serve the condenser_api endpoints publicly again — it's one of the more reliable endpoints for dev work.