You can use ODBC/JDBC (for JDBC, use the one from CDH 4.6) to issue any SQL
stmt to Impala, including insert.
For your INSERT query, the INSERT is distributed across all impalads. If
you look at the plan, each plan fragment will do some reading and then
write to test3.
Here's the link to our documentation:
http://www.cloudera.com/content/cloudera-content/cloudera-docs/Impala/latest/Installing-and-Using-Impala/Installing-and-Using-Impala.html
On Mon, Mar 10, 2014 at 7:54 PM, wrote:
Thanks alan.
Other Question:
1: Can i user odbc/jdbc insert data to impala table (use parquet);
2: I test in impala-shell, once i execute "insert into table test3 select
* from test2;"
(test2 have 3 rows) Test3 add 3 rows,but why test3 hdfs directory add
3 files,and file is so small?
在 2014年3月11日星期二UTC+8上午10时27分36秒，Alan写道：
To unsubscribe from this group and stop receiving emails from it, send an email to impala-user+unsubscribe@cloudera.org.Thanks alan.
Other Question:
1: Can i user odbc/jdbc insert data to impala table (use parquet);
2: I test in impala-shell, once i execute "insert into table test3 select
* from test2;"
(test2 have 3 rows) Test3 add 3 rows,but why test3 hdfs directory add
3 files,and file is so small?
在 2014年3月11日星期二UTC+8上午10时27分36秒，Alan写道：
Hi,
Unfortunately, there's no easy way to d it easily. The recommended way is
to put the file in HDFS first, and then use Hive or Impala to do the
conversion.
Parquet requires that the whole file stay in a block. Using "hadoop fs
-put" might violate this condition.
Thanks,
Alan
an email to impala-user+unsubscribe@cloudera.org.Unfortunately, there's no easy way to d it easily. The recommended way is
to put the file in HDFS first, and then use Hive or Impala to do the
conversion.
Parquet requires that the whole file stay in a block. Using "hadoop fs
-put" might violate this condition.
Thanks,
Alan
On Mon, Mar 10, 2014 at 7:14 PM, wrote:
Hi,
Customer can only provide csv format file by time period. I plan use
impala:
1: Convert csv file to parquet file directly out of hadoop cluster;
2: Then put file to hdfs specific directory.
3: Alter table add a new table's partions (by time period)
4: All is OK
Now How can i convert csv file to parquet file directly?
To unsubscribe from this group and stop receiving emails from it, send
an email to impala-user...@cloudera.org.
To unsubscribe from this group and stop receiving emails from it, sendHi,
Customer can only provide csv format file by time period. I plan use
impala:
1: Convert csv file to parquet file directly out of hadoop cluster;
2: Then put file to hdfs specific directory.
3: Alter table add a new table's partions (by time period)
4: All is OK
Now How can i convert csv file to parquet file directly?
To unsubscribe from this group and stop receiving emails from it, send
an email to impala-user...@cloudera.org.
