Consider the following setup: there are two sites, A with 4 nodes, and B with 2 nodes. Having a CACHE which is backed up to B site.
Size of the data and given memory in site B nodes allows for a distributed cache with numowners==1 and a single node cannot contain the whole data. (won't fit into its memory)
Some clients are directed exclusively to site B (they are local to site B). If one of the nodes fails in site B, what happens when an entry is requested which is not cached in the still alive node?
Will the request be served transparently by getting the requested data from the A site's nodes?
More generally: is cross-datacenter replication suitable for such uses? E.g. use backup sites to bring the cache to clients where they can access it locally?
Retrieving data ...