npub1y0emt2wlpsezcnmxtyrpf33qe7gwy5u8yzssvv6uw53em0k32t7q7smm9n (npub1y0e…mm9n) we’ve got background compaction in CouchDB paired with an IO scheduler, it makes it a non-entity in terms of operational performance, so aid say it is a viable option.
But I understand it is not the absolute most efficient way to do this. Also, we have very low hanging fruit in the compaction process itself.