Which statement describes Distributed Cache behavior?

Which statement describes Distributed Cache behavior?

• A. Cache data is replicated on each node
• B. Cache data is evenly distributed across all Confluence nodes in a cluster
• C. Cache data is invalidated remotely by other nodes
• D. Cache data is stored on a single node

Answer: (B)

The key idea here is how a distributed cache spreads responsibility across multiple nodes. In a distributed cache, data is partitioned so that different pieces of the cache live on different nodes, typically using hashing or a similar mechanism to map keys to nodes. This means the cache load is balanced and no single node becomes a bottleneck, which is essential for scaling out in a cluster like Confluence. Therefore, describing cache data as evenly distributed across all nodes in the cluster best captures this behavior. It explains why performance scales with added nodes and why memory is used more efficiently overall. If the cache were replicated on every node, memory usage would balloon and updates would require more coordination. If all data were stored on a single node, you’d face a single point of failure and a bottleneck. If data were merely invalidated remotely, that speaks to coherence or invalidation methods rather than how the data is distributed in the cluster.

The key idea here is how a distributed cache spreads responsibility across multiple nodes. In a distributed cache, data is partitioned so that different pieces of the cache live on different nodes, typically using hashing or a similar mechanism to map keys to nodes. This means the cache load is balanced and no single node becomes a bottleneck, which is essential for scaling out in a cluster like Confluence.

Therefore, describing cache data as evenly distributed across all nodes in the cluster best captures this behavior. It explains why performance scales with added nodes and why memory is used more efficiently overall.

If the cache were replicated on every node, memory usage would balloon and updates would require more coordination. If all data were stored on a single node, you’d face a single point of failure and a bottleneck. If data were merely invalidated remotely, that speaks to coherence or invalidation methods rather than how the data is distributed in the cluster.