If there are many tombstone rows, Seek()/Prev()/Next() may take (potentially much) longer time, because they have to scan many tombstone rows. This is RocksDB internal issue and MyRocks is affected by the limitation. We need find ways to fix -- either fixing RocksDB itself or making workarounds in MyRocks side. The latter will likely to make automatic and adaptive compaction thread, based on tombstone row statistics.
If there are many tombstone rows, Seek()/Prev()/Next() may take (potentially much) longer time, because they have to scan many tombstone rows. This is RocksDB internal issue and MyRocks is affected by the limitation. We need find ways to fix -- either fixing RocksDB itself or making workarounds in MyRocks side. The latter will likely to make automatic and adaptive compaction thread, based on tombstone row statistics.