Cloudera Impala

• To write a table with field names that total more than 4000 characters, use CSV instead of Avro.
• When creating field names and writing queries in Impala, note that identifiers (names of databases, tables or columns) in Impala have certain restrictions. While reserved keywords are supported, other identifiers are not supported. For example:
  • Identifiers cannot have a space.
  • Identifiers cannot begin with a number (e.g., 1st is invalid)
  • Identifiers can only be alphanumeric (e.g., no percent sign (tax%), dollar sign (money$), symbols (>, < etc) or any other non-ASCII characters.
  • Identifiers are case insensitive (e.g., T1 and t1 are the same identifier)

• Quoted fields in table styles are not supported.
• Unicode characters are not provided.

Delete and Append statements are not supported.

• Strings are limited to 32,767 bytes.
• The TINYINT data type range is -128 to 127 and therefore needs to be mapped to INT16.

• Due to an Impala database limitation, you cannot select a table with the Connect In-DB Tool and then write to HDFS Avro with either the Overwrite Table (Drop) or Create New Table Creation Mode in the Write Data In-DB Tool.