Has anybody written the RocksDBAggregationRepository yet, or is there a way to make the LevelDBAggregationRepository use RocksDB? I was trying to go from in memory aggregation of a large json document dataset to a leveldb based one so that memory use would be limited, but the performance of the leveldb repo was around 10X slower than in memory. I am looking for a very fast persistent aggregation repository that uses memory efficiently. If I were to use a rocksdb agg repository on enterprise internal SSD's I'm wondering how much faster it would be than the leveldb one.
Anybody have any thoughts?
