HBase and/or Cassandra
I have a transactional system (federated ESB clusters) that process
millions transactions daily. I need a reliable, consistent and partition
tolerance database system.
I am thinking of logging realtime transactions into Cassandra (as I
understand it has a faster write capacity) and archiving these
transactions to HBase for query/analyses operations.
Have anyone seen this type of architecture before? Is it a
recommended/scalable/practicable direction?
I will also like suggestions on the best architecture to solve this problem.
Note that data being stored are large XML/JSON data, and sometimes
structure-less (free flow text data). Yet, I need to store them in a
manner that can be queried for monitoring/debugging and pulled up for
reporting as well.
Regards.
No comments:
Post a Comment