When we use aqe, we need use mapId to filter the data which we don't need, If we sort MapId before the data are flushed. We split the data to segments, if a segment don't have the data which we want to read, we will drop the data. If data is sorted by mapId, we can filter more data and mprove our performance.