Which sharding approach offers maximum flexibility for isolating hot users to dedicated shards but may introduce latency due to directory lookups?

Which sharding approach offers maximum flexibility for isolating hot users to dedicated shards but may introduce latency due to directory lookups?

• A. Range-Based Sharding
• B. Consistent Hashing
• C. Directory-Based Sharding
• D. Hot Spot Problem

Answer: (C)

Directory-based sharding uses a central mapping that explicitly assigns individual keys (such as hot users) to specific shards. This gives the most control to isolate hot users on dedicated shards, because you can pin those keys to their own shards and reassign them quickly without changing the overall shard boundaries or redistributing large swaths of data. That flexibility is valuable when certain users or workload patterns need strong separation to prevent overload on common shards. However, routing data to the correct shard depends on looking up the directory for the mapping before every operation. Those directory lookups introduce additional network hops and potential latency, especially if the directory becomes a bottleneck or must be queried for every read or write. Caching can mitigate this, but cache misses and directory availability still pose latency and reliability concerns. Range-based sharding and consistent hashing aim to distribute keys efficiently without per-key mapping to a single directory, which reduces routing overhead but makes pinpointing and isolating specific hot users more cumbersome or disruptive to the overall shard balance. The hot spot problem describes a symptom where certain keys overload a shard, not a sharding mechanism itself.

Directory-based sharding uses a central mapping that explicitly assigns individual keys (such as hot users) to specific shards. This gives the most control to isolate hot users on dedicated shards, because you can pin those keys to their own shards and reassign them quickly without changing the overall shard boundaries or redistributing large swaths of data. That flexibility is valuable when certain users or workload patterns need strong separation to prevent overload on common shards.

However, routing data to the correct shard depends on looking up the directory for the mapping before every operation. Those directory lookups introduce additional network hops and potential latency, especially if the directory becomes a bottleneck or must be queried for every read or write. Caching can mitigate this, but cache misses and directory availability still pose latency and reliability concerns.

Range-based sharding and consistent hashing aim to distribute keys efficiently without per-key mapping to a single directory, which reduces routing overhead but makes pinpointing and isolating specific hot users more cumbersome or disruptive to the overall shard balance. The hot spot problem describes a symptom where certain keys overload a shard, not a sharding mechanism itself.