What term describes a shard that receives disproportionately more traffic than others, undermining sharding benefits?

What term describes a shard that receives disproportionately more traffic than others, undermining sharding benefits?

• A. Hot Spot Problem
• B. Cross-Shard Queries
• C. Saga Pattern
• D. Two-Phase Commit

Answer: (A)

Sharding aims to spread load evenly across multiple shards, but when one shard ends up handling a disproportionate share of traffic, you’re dealing with a hot spot. This undermines the whole point of sharding because that single shard becomes the bottleneck, limiting overall system throughput and driving up latency even if the other shards are underutilized. Hot spots usually arise from skewed data distributions or access patterns—think a few popular keys or items that attract much more traffic than the rest, or predictable time-based bursts. You’ll notice signs like a single node showing high CPU or memory pressure, growing backlogs, and elevated tail latency while other shards stay relatively idle. Mitigations focus on distributing load more evenly: using a different partitioning scheme such as hash-based or consistent hashing to avoid concentrating hot keys on one shard; splitting hot keys into finer-grained subkeys or adding more shards and rebalance data; caching hot data or using read replicas to take excess read pressure; and sometimes routing strategies that isolate or spread out traffic for the busiest keys. Terms like cross-shard queries describe operations that touch multiple shards rather than a single shard carrying extra load; the Saga pattern and Two-Phase Commit are about coordinating distributed transactions, not load distribution.

Sharding aims to spread load evenly across multiple shards, but when one shard ends up handling a disproportionate share of traffic, you’re dealing with a hot spot. This undermines the whole point of sharding because that single shard becomes the bottleneck, limiting overall system throughput and driving up latency even if the other shards are underutilized. Hot spots usually arise from skewed data distributions or access patterns—think a few popular keys or items that attract much more traffic than the rest, or predictable time-based bursts. You’ll notice signs like a single node showing high CPU or memory pressure, growing backlogs, and elevated tail latency while other shards stay relatively idle.

Mitigations focus on distributing load more evenly: using a different partitioning scheme such as hash-based or consistent hashing to avoid concentrating hot keys on one shard; splitting hot keys into finer-grained subkeys or adding more shards and rebalance data; caching hot data or using read replicas to take excess read pressure; and sometimes routing strategies that isolate or spread out traffic for the busiest keys.

Terms like cross-shard queries describe operations that touch multiple shards rather than a single shard carrying extra load; the Saga pattern and Two-Phase Commit are about coordinating distributed transactions, not load distribution.