Which statement about Directory-Based Sharding is true?

Which statement about Directory-Based Sharding is true?

• A. It has no latency due to directory lookups.
• B. It introduces latency due to directory lookups and can be a single point of failure.
• C. It makes cross-shard transactions trivial.
• D. It cannot move records between shards.

Answer: (B)

Directory-based sharding uses a central (or widely used) directory that maps each data item to the shard that stores it. Because every read or write must first query this directory to know where to route the request, you incur an extra lookup overhead. If that directory is a single centralized service, it also becomes a single point of failure, so its outage or slowdown can impact the whole system. This approach does not eliminate latency; it adds it through lookups. It also does not make cross-shard transactions trivial—these remain more complex because coordinating work across multiple shards is still needed. And moving records between shards is possible, though it requires updating the directory and transferring the data accordingly, not an inherent impossibility. So the statement that best captures the reality is that directory lookups introduce latency and the directory can be a single point of failure.

Directory-based sharding uses a central (or widely used) directory that maps each data item to the shard that stores it. Because every read or write must first query this directory to know where to route the request, you incur an extra lookup overhead. If that directory is a single centralized service, it also becomes a single point of failure, so its outage or slowdown can impact the whole system.

This approach does not eliminate latency; it adds it through lookups. It also does not make cross-shard transactions trivial—these remain more complex because coordinating work across multiple shards is still needed. And moving records between shards is possible, though it requires updating the directory and transferring the data accordingly, not an inherent impossibility.

So the statement that best captures the reality is that directory lookups introduce latency and the directory can be a single point of failure.