Hermes Wiki

Sharding

Splitting a single logical database across multiple physical nodes for scale — the DB/infra-configuration side of what ConsistencyConcurrency/Partitioning-Sharding covers conceptually.

Why we need this / what value this brings

Lets a dataset and its write throughput scale past what a single database node can handle.

When to use this

Only once vertical scaling (bigger instance) and read replicas are proven insufficient — sharding adds major operational and query complexity.

How to use or implement this

Postgres's built-in declarative partitioning (splitting one big table across partitions on one instance) is a much simpler first step than true multi-node sharding.

Research questions

  • Postgres-native partitioning (declarative partitioning on one instance) vs true multi-node sharding — very different complexity levels; which would ever apply to Localz first?

Empty folder — drop notes, links, and findings here as you research.

Hermes Wiki