triplesheep commented on issue #25556: [SPARK-28853][SQL] Support conf to 
organize file partitions by file path
URL: https://github.com/apache/spark/pull/25556#issuecomment-524783325
 
 
   Hi, @cloud-fan :) we can use partitionBy to do this but I don't do that for 
two reasons:
   1. partitionBy will be expensive because of shuffle
   2. I don't want all data in a partition ends in the same RDD partition 
because it may be very large and the write may be slow. It seems like data 
skew. When use partitionBy it will put same key into same RDD partition and 
will encounter the problem I said above.
   
   I want one RDD partition only contains the data from one data partition 
doesn't means I want data in same data partitions locates into the same RDD 
partition. For example:
   Data Distribution:
   day = 1: file1  file2 file3 file4
   day = 2: file5
   day = 3: file6
   What I want in RDD partition(split by conf FILES_MAX_PARTITION_BYTES):
   RDD partition1: file1 file2   (day=1)
   RDD partition2: file3 file4 (day=1)
   RDD partition3: file5 (day=2)
   RDD partition4: file6 (dat-3)

----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
[email protected]


With regards,
Apache Git Services

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to