Hive ORC compression

Question

I run following code in hive v0.12.0 and I expect to get three tables compressed using different methods and therefore size and content of the files should be different.

--- Create table and compress it with ZLIB
create table zzz_test_szlib
  stored as orc
  tblproperties ("orc.compress"="ZLIB")
  as
select * from uk_pers_dev.orc_dib_trans limit 100000000;

--- Create table and compress it with SNAPPY
create table zzz_test_ssnap
  stored as orc
  tblproperties ("orc.compress"="SNAPPY")
  as
select * from uk_pers_dev.orc_dib_trans limit 100000000;

--- Create table and DO NOT compress it
create table zzz_test_snone
  stored as orc
  tblproperties ("orc.compress"="NONE")
  as
select * from uk_pers_dev.orc_dib_trans limit 100000000;

When I check the tables metadata using describe or through Hue I get:

Name             Value                                            Value                                            Value
---------------- ------------------------------------------------ ------------------------------------------------ ------------------------------------------------
tableName        test_orc_zlib                                    test_orc_snappy                                  test_orc_none
location:hdfs    /user/hive/warehouse/test_orc_zlib               /user/hive/warehouse/test_orc_snappy             /user/hive/warehouse/test_orc_none
inputFormat      org.apache.hadoop.hive.ql.io.orc.OrcInputFormat  org.apache.hadoop.hive.ql.io.orc.OrcInputFormat  org.apache.hadoop.hive.ql.io.orc.OrcInputFormat
outputFormat     org.apache.hadoop.hive.ql.io.orc.OrcOutputFormat org.apache.hadoop.hive.ql.io.orc.OrcOutputFormat org.apache.hadoop.hive.ql.io.orc.OrcOutputFormat
compressed       FALSE                                            FALSE                                            FALSE
serializationLib org.apache.hadoop.hive.ql.io.orc.OrcSerde        org.apache.hadoop.hive.ql.io.orc.OrcSerde        org.apache.hadoop.hive.ql.io.orc.OrcSerde
orc.compress     ZLIB                                             SNAPPY                                           NONE
numFiles         1                                                1                                                1
totalSize        289970088                                        289970088                                        289970088
tableType        MANAGED_TABLE                                    MANAGED_TABLE                                    MANAGED_TABLE

In the metadata it shows compressed=FALSE, but I don’t know how to change this and how this will affect.

But if I compare table’s data they all binary identical.

[~]$ hadoop fs -ls /user/hive/warehouse/test_orc_*
-rw-r--r--   3 andrey supergroup  289970088 2014-05-07 13:19 /user/hive/warehouse/test_orc_none/000000_0
-rw-r--r--   3 andrey supergroup  289970088 2014-05-07 12:34 /user/hive/warehouse/test_orc_snappy/000000_0
-rw-r--r--   3 andrey supergroup  289970088 2014-05-07 11:48 /user/hive/warehouse/test_orc_zlib/000000_0

I tried to change/remove these options, but it makes no difference:

SET hive.exec.compress.intermediate=true;
SET hive.exec.compress.output=true;
SET mapred.output.compression.type=BLOCK;

Also I tried to use different source table (stored as TEXTFILE), no difference.

Any thoughts or suggestions?

Answer 1

I believe this is due to a known bug in 0.12
Have a look at this Jira HIVE-6083