This issue is primarily concerning usability.

Both Innodb and Rocksdb have *_flush_at_trx_commit with same options.

This got me thinking, we should align more as it makes it easier to switch engines without as big a learning curve.

Example:

rocksdb_buffer_pool_size could replace rocksdb_block_cache_size

alternatively, both rocksdb_block_cache_size and innodb_buffer_pool_size could be replaced with something like rocksdb_cache_size and innodb_cache_size.. could go further with *_cache_instances _log_file_size etc