Versions Compared

Old Version 10

changes.mady.by.user Senhong Liu

Saved on

compared with

New Version 11

changes.mady.by.user Senhong Liu

Saved on

Key

  • This line was added.
  • This line was removed.
  • Formatting was changed.

...

  • The input source stream is already partitioned in the external system.
    • For example, user data have already been grouped by the user ID in the external system.
    • A use case in the user ML is case 1.
  • The DataStream is already grouped by key previously with the keyBy().
    • For example, a record with three fields, (key, val_1, val_2), can be grouped first by the key with the keyBy() operation. Then, the data can be summed with val_1 and val_2, separately. 
    • A use case in the user ML is case 12.

However, one limitation is that for the first use case, if the input source stream is NOT partitioned exactly the same way Flink's keyBy would partition the data, the reinterpretation would cause the runtime error since KeyGroup mismatch.

...