Problem
Single database hits limits. How to split data across nodes? Wrong strategy = hotspots, impossible resharding, cross-shard queries.
Also known as: sharding-strategies, horizontal-partitioning, data-partitioning, consistent-hashing
Choose and implement partitioning strategy: hash, range, directory, consistent hashing — trade-offs for distribution, resharding, and query routing.
Single database hits limits. How to split data across nodes? Wrong strategy = hotspots, impossible resharding, cross-shard queries.
High — more nodes, coordination
High — resharding, monitoring per shard
High — strategy choice, query routing, resharding
Hot shard: 80% load on one shard
Resharding stall: backfill falls behind, switch delayed
Cross-shard transaction: partial commit, inconsistency
Shard count too high: metadata overhead, slow planning
Routing layer SPOF: proxy down = all shards unreachable