For example, with typical page sizes of 8 KB or 16 KB, in-page processing is inherently sequential and less CPU-efficient than random data access.
例如,对于典型的8 KB或16 KB的页面,页内处理天性是连续的,CPU效率低于随机数据访问。
2
One of the key design concepts behind Percolator was to provide randomaccess to the repository to process documents individually, and thus avoiding the global processing required by MapReduce.