Hitachi

In-Memory Data Grid Hitachi Elastic Application Data Store


2.5.3 Adding EADS servers (scaling out) and distributing data

EADS enables you to add new EADS servers to a cluster without having to stop the cluster. This is called scale-out processing. You can add an EADS server at any desired location on the consistent hashing circumference.

When scale-out processing is performed, the range within which a new EADS server is added is divided. The range containing values that are greater than the value at the location of the added EADS server (hash value) becomes the new range. The EADS server that had been managing the previous range transfers the newly added range of data to the added EADS server.

Adding new EADS servers to a cluster enables you to increase the overall size of the physical memory and improve performance in the entire cluster. By dividing a range, you can reduce the memory usage by the EADS server that had been managing the previous range and reduce its workload.

The following figure provides an overview of adding one new EADS server to a cluster.

Figure 2‒18: Overview of scale-out processing (adding an EADS server)

[Figure]