org.openspaces.grid.esm
Interface ElasticScaleHandler

All Known Implementing Classes:

NullElasticScaleHandler

public interface ElasticScaleHandler

An API to be called upon when in need to scale out to a different machine, or scale in when the machine is no longer needed. When scaling

 Disclaimer: This interface and the elastic data grid functionality is provided as a technology preview in XAP 7.1. 
 As such, it is subject to API and behavior changes in the next XAP releases without going through the usual deprecation process 
 of the XAP API.
 

Method Summary

boolean	accept(Machine machine) A machine can be accepted/not-accepted (filtered) for whatever reason at any point in time.
void	init(ElasticScaleHandlerConfig config) An initialization call, parameterized with the configuration used at deployment.
void	scaleIn(Machine machine) A scale in request to terminate a machine.
void	scaleOut(ElasticScaleHandlerContext context) A scale out request to start a machine.

Method Detail

init

void init(ElasticScaleHandlerConfig config)

An initialization call, parameterized with the configuration used at deployment.

Parameters:

config - The scale configuration.

accept

boolean accept(Machine machine)

A machine can be accepted/not-accepted (filtered) for whatever reason at any point in time. This will prevent from a GSC to be started on it.

Parameters:

machine - The machine to accept to filter.

Returns:

true to use this machine to start a GSC on; false to skip it.

scaleOut

void scaleOut(ElasticScaleHandlerContext context)

A scale out request to start a machine. Implementation may choose to block until it has scaled out (which will cause the ESM to wait until the call has returned, preventing it from handling other processing units). It's advised to scale out asynchronously due to long startup delays. As long as the machine has not been discovered, the ESM will repeatedly call this method until a new machine has been allocated. It's up to the implementation class to maintain the state and not allocate new resources before the pending allocations have completed. When a new machine is started, it should start a Grid Service Agent (GSA).

Parameters:

context - Context details

scaleIn

void scaleIn(Machine machine)

A scale in request to terminate a machine. Implementation may choose to ignore it, or decide to terminate considering cost/benefit factors. Will be called only once per-last GSC on this machine to terminate. Implementation may choose to block until it has scaled in (which will cause the ESM to wait until the call has returned, preventing it from handling other processing units).

Parameters:

machine - The machine to terminate.