Infrastructure · Data Structures
Shard
Deterministic partitioning of state
Definition
Mechanism
The fundamental primitive of distribution. It deterministically partitions a Resource, State, or Vector into disjoint subsets (Shards) based on a Key (e.g., ID, Hash, Time). Unlike Decompose, which splits complexity, Shard splits volume and load.
Invariants
- Conservation: Sum(Shards) == Total.
- Disjointness: Intersection(Shard_A, Shard_B) == Empty.
- Determinism: Key K always routes to Shard S.
Risks
Failure modes
Hot shard
skewed keys concentrate load on one partition — a consequence of Determinism.
Rebinding on resize
changing the partition count invalidates every routing decision taken under the old count.
Cross-shard operation
spanning shards requires coordination, because Disjointness guarantees no shard holds another's data.
Relationships
Builds on
References
Physics
An invariant quantity preserved under a closed system's transformations — a substrate law
Mind
Dividing complexity into manageable units
Finite allocatable entity
System condition at time T
Semantic position array
Used by
References
Mind
Agent bifurcation on context saturation: active vs archival specialists
Society
Interlocking state threads
Reference
Parameters
key_functionEnum, {Hash, Range, Directory}- Partitioning strategy for distributing data across shards
partitionsInteger, [2, 256]- Number of partitions to divide state into
Data schema
{
"type": "object",
"required": [
"shard_id",
"partition_index"
],
"properties": {
"shard_id": {
"type": "string",
"description": "Unique identifier for this shard"
},
"key": {
"type": "string",
"description": "The key used to route to this shard"
},
"partition_index": {
"type": "integer",
"description": "Which partition this shard belongs to"
},
"data": {
"description": "The shard's contents"
}
}
}