NoSQL Training Online

Cassandra commits in two steps – first to log on disk and then to memtable (which is an in-memory structure).

They are created per column famiy.

Impedance mismatch refers to the two separate representations for translation due to the difference between the relational model and the in-memory data structures. This is so because in the relational data model, data has to be simple because it has to be organized into tables and rows, but in the in-memory data structures, data can be much richer, therefore has to be translated into a relational representation for storing on a disk.

There are two major differences between replication and sharding:

• Replication takes the same data and put it on various nodes. Sharding puts different data on different nodes.
• Replication enhances only read performance but does not much affect write performance. Sharding greatly improves read and write performance.

It is an open-source distributed database management system from Apache that is highly scalable.

• Cluster
• Keyspace
• Column
• Column family

Cloudera developed an open-source software program called Flume which aggregates and moves huge amounts of data around a Hadoop cluster.

• Single node
• Pseudo-distributed
• Fully distributed