|
Hi,
I believe Kryo is only use during RDD serialization (i.e. communication between nodes), not for saving. If you want to compress output, you can use GZip or snappy codec like that : val codec = "org.apache.hadoop.io.compress.SnappyCodec" // for snappy val codec = "org.apache.hadoop.io.compress.GzipCodec" // for gzip System.setProperty("spark.hadoop.mapreduce.output.fileoutputformat.compress", "true") System.setProperty("spark.hadoop.mapreduce.output.fileoutputformat.compress.codec", codec) System.setProperty("spark.hadoop.mapreduce.output.fileoutputformat.compress.type", "BLOCK") (That's for HDP2, for HDP1, the keys are different) Regards Guillaume
--
|
- Turning kryo on does not decrease binary output Aureliano Buendia
- Re: Turning kryo on does not decrease binary output Guillaume Pitel
- Re: Turning kryo on does not decrease binary ou... Aureliano Buendia
- Re: Turning kryo on does not decrease binar... Andrew Ash
- Re: Turning kryo on does not decrease b... Aureliano Buendia
- Re: Turning kryo on does not decrease binar... Guillaume Pitel
- Re: Turning kryo on does not decrease b... Aureliano Buendia
- Re: Turning kryo on does not decre... Guillaume Pitel
- Re: Turning kryo on does not d... Aureliano Buendia
- Re: Turning kryo on does n... Andrew Ash
- Re: Turning kryo on does n... Aureliano Buendia
- Re: Turning kryo on does n... Guillaume Pitel

