How sharding is supported in infinispan?

[udit-mishra-5113a21a]

1.) How does infinispan supports scaling out? For scaling out, adding a node with same configurations as of participant nodes of cluster and starting it will be sufficient, they will form a topology, right?

2.) How does infinispan supports scaling up? As something I know is giving a JVM heap size and max size parameters, does it has to do anything with scaling up, like if I need more memory, I will stop all nodes, change the JVM parameters and start them again, will that do?? I also want to know what should be the recommended values for heap size and max size??

[galder.zamarreno]

Scaling out is provided via Infinispan's clustering facilities which are powered by JGroups, a group communication layer. Consistent hash distribution and total replication are the two main mechanisms to ship data around, and yes, dynamic topology changes are supported.

Yeah, scaling up to provide different JVM settings requires restart. Data backups can be configured via shared cache store configurations, or rolling upgrades can be performed to do live upgrading.

Heap sizing is heavily dependant on your data.