Background:

Have 2 Flume JVMs with an Avro connection between them.  They are moving
syslog data to Accumulo.

 

Now I wish to pick up some MIB data and also store through Flume to the
same repository in another table.

 

I am looking for some recommendations for gathering MIB data into the
Flume system.

 

Should I write a Java client to pull the MIB, format the data, and push
the data into a Flume source? We have a JSONSource patterned after the
AvroSource.

 

Or would it be better to use ExecSource? But, that is for non-Java
calls?

 

Should the Java MIB reader live inside of Flume - be a Flume component?
Or should the Java MIB reader be a Flume client?

 

 

 

Thanx in advance for any insights, advice, and recommendations.

Reply via email to