Concepts

• it is nosql but does not have database
• uses something called index objects to refer as collection (mongodb)
• scale wise uses multiple indices

Cluster
Database

Types & Indices
types is the key, index is the object

eg. index twitter with two types user and tweet
or multiindex user as index with two types info and tweet

subdivide index into pieces called shards

• each shard can be hosted on any node in the cluster
• horizontal volumn split, good for scale and good for parallizelize operations performance

replicas - replications of shards
since failure is expected in network enviornment, failover mechanism to allow one to make copies of shards

• good for failover mechanism
• good for scaling volume since it is basically shards

complete replica (all shards are copied)

primary shards - original shards
by default one index 5 primary shards and 1 complete replica

Scaling
limit shards replicas, more index objects