The best practices for Storm/Cassandra applications
When working with distributed applications that have SLAs operating 24/7 with a very high velocity and a miniscule average processing time, certain aspects become extremely crucial to be taken care of:
Network latency plays a big role in real-time applications and can make or break products, so make a very informed and conscious decision on the placement of various nodes in a data center or across data centers. In such situations, it's generally advisable to keep ping latency at a minimum.
The replication factor should be around three for Cassandra.
Compaction should be part of routine Cassandra maintenance.
