Memory Considerations

Memory is the most restrictive element in Apache HBase. Performance-tuning techniques are focused on optimizing memory consumption.

From a schema design perspective, it is important to bear in mind that every cell stores its value as fully qualified with its full row key, column family, column name, and timestamp on disk. If row and column names are long, the cell value coordinates might become very large and take up more of the Apache HBase allotted memory. This can cause severe performance implications, especially if the dataset is large.

Tip

Keep the number of column families small to improve performance and reduce the costs associated with maintaining HFiles on disk.