Hi,
Flume is not limitted to long data.  It has a flume-source to observe 
directories and can load existing files. The morphlines from kite-sdk can help 
you even transform data on the fly. 
Cheers,
Mirko



Von Samsung Mobile gesendet

<div>-------- Ursprüngliche Nachricht --------</div><div>Von: Shashidhar Rao 
<[email protected]> </div><div>Datum:06.06.2014  16:26  (GMT+01:00) 
</div><div>An: [email protected] </div><div>Betreff: Best way to transfer 
data to HDFS </div><div>
</div>Hi Hadoop Experts,

I have to transfer 5 yrs of historical billing data to the tune of 25 -30 TB to 
HDFS to be later analyzed by Map Reduce program. Sqoop is out of question as 
these files are not residing in OLTP and so is Flume as these are not log files 
generated by App Server.

What tools are available to transfer such Big data? Would it be prudent to use 
Hadoop put command to transfer these files

Regards
Shashi

Reply via email to