2.5.1 Overview of data distribution

This subsection provides an overview of data distribution.

Figure 2‒17: Data distribution by consistent hashing

In consistent hashing, the servers and keys are treated as being placed on the same circumference. Sequential integers are assigned in ascending order to this circumference in a counterclockwise direction.

In this figure, first, the hash values of the EADS servers (from 1 through 5) and the keys for the data to be stored are obtained. Then, the servers and keys are placed on the circumference according to their hash values.

The key for each data item is then stored on the first EADS server located clockwise from the location at which the key was placed.

A range of hash values obtained by separating the consistent hashing circumference in a cache by the location of each EADS server is called a range. Ranges are managed by range IDs. In this figure, a key placed in range 4 is stored in EADS server 4.

This circle has nothing to do with the physical placement of the EADS servers.

To Page Top