Scaling elastically or Auto Scaling with built-in fault tolerance

Before we go into Auto Scaling we need to give a microscopic view of how HBase accomplishes auto-sharding and how the distributed components within HBase architecture work.

Let's first look at Region.

HBase Regions are a subgroup of table's data which is adjoining; these are in a sorted order of range of rows which sit together, and these regions are distributed across the clusters. Region never overlaps with other regions and the job of a single region server is to serve to expose Region details to the region client at any given point of time; this way HBase provides a guarantee a very strong consistency.

The region has many stores.

A stores hosts a MemStore and deals with the store files (commonly known as HFiles). MemStore is an In-Memory state of the data and takes care of the modifications of the key/value pairs. Due to the following reason (as following) when a flush process is initiated the data which is residing in MemStore...