Assuming that I want to setup a infinspan cluster topology consisting of the following pieces

• regional read-only consumers which should work with the cluster cache members in the regional dc's only
• cluster members are distributed on a global level, i.e. to regional dc's in APAC, Europe, USA in order to allow dedicated cache updating applications to put regional values into the cache which will be then distributed to all global instances
• assume that we have a distributed caching strategy

What would be the best strategy for such a setup

• is there a way to get that solved with one cluster group or are multiple "regional" cluster group (APAC, EU, USA) required with some updater process per region which are connected to each regional cluster group and which will propagate regional data changes to each member ?
• Is there a way to make the client regional aware, i.e to tie its lookup to the nodes which are within its region (equal latency) and to filter out the global dispersed cache nodes (inter contintental latency)