We have a Cassandra sink that writes to different column families based on the "event type" (a header attribute). Each type has its own "handler". Right now the handlers are hard-coded, but I think your idea of using a configuration file would allow it to be generic. We also write to multiple indexes, each of which could also be defined by configuration.
I believe a generic Cassandra sink that can be configured this way would be a good addition.
On Feb 11, 2013, at 2:08 AM, Sri Ramya wrote:
> Can you please explain what is the present data model of flume in
> cassandra. Can we change it by configuration file.
> i am tring to devolop new sink for the flume cassandra data model.
> please tell me...