Cassandra and HBase

1. What is the difference between Cassandra and HBase

Origin of both these NoSQL's is Google's BigTable.

HBase sits on HDFS and has tight dependency on it. Cassandra is  

HBase is Master Slave Architecture. Cassandra is not.

Apache Phoenix for HBase and nothing for Cassandra

Cassandra excellent for single row read performance or selecting multiple rows based on Column-Value Index.

Cassandra has better scalability when compared with HBase.

//////////////////////////////////////////////////////////////////////////////

Cassandra:

• Object storage model.

• Simple setup, low maintenance. Scale up, scale down by adding or removing nodes is simple. No need to worry about re-syncing, balancing of data. These are automatically done. Basically less Administration work.

• Cluster can be scaled in various geographical locations across the world. Commonly used to create a database that is spread across nodes in more than one data centre, for high availability.

• Good in case of  high velocity of Random Reads & Writes. Columnar, distributed and de-centralized architecture.

• Less consistant when compared with HBase. Highly Available. 
• Consistency and Latency are inversely proportional in cassandra, When full consistency is set i.e multiple replications. In this case a write operation could complete only after writing to all the replicas which in turn increases Latency. So, Tradeoff between consistency and latency.

• Flexible schema where columns can be increased as per requirements.

• Good performance with non-group by operations. No Multiple Secondary Index needs.

HBase:

• Block storage

• Optimized for Reads

• Range based scans because of block storage as like Hadoop. Range is bigger as the storage this is Block storage. If the query range falls in a single block then reading is very fast. 

• Strong Consistency

• Doesn't have Geographical locations cluster footprint

• Facebook messenger used HBase.

Reference: https://www.youtube.com/watch?v=nIPPTn4IC2Y&t=55s
https://www.youtube.com/watch?v=zCOFC8IVn3Q&t=18s