[HBASE-16818] Avoid multiple copies of binary data during the conversion from Result to Row

• Type: Improvement
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 2.0.0
• Component/s: spark
• Hadoop Flags: Reviewed
• Assignee: Weiqing Yang
• Reporter: Weiqing Yang
• Created: 12/Oct/16 22:22
• Updated: 06/Apr/18 04:58
• Resolved: 14/Oct/16 17:17

Description

In the buildRow() of HBaseRelation, CellUtil.cloneValue will already create a copy of the data. If the data type is BinaryType, another copy is being made within Utils.hbaseFieldToScalaType in Utils.scala. Generally, binary data can be fairly large, so copying may be an expensive operation.

Issue Links

• links to: ReviewBoard