Small table considerations

Small tables present a challenge when partitioning.

By: Balazs Jeszenszky, Field Engineer, Cloudera, Inc.

Partitioning is often suggested as a way to speed up any query. As seen in earlier sections, this may not be true for small tables or ineffective partitioning. Instead, consider using HDFS caching for heavily queried small tables. Storing the data in the DataNode's cache greatly speeds up scans at the cost of memory allocation on the DataNodes: