Storage

Storage in NYXDB is governed by named policies that bind a table to durability, delta flush behavior, index placement, part movement, WAL, and compaction. See the storage model concept for the delta → parts → compaction lifecycle.

Selecting a policy

CREATE TABLE cold (id UInt64, payload String, PRIMARY KEY (id))
  SETTINGS storage_policy = 'disk_data';

Policies observed in the engine include memory_data, disk_data, default, tiered, and aged_tiered.

Inspecting storage

Configured policies:

SELECT name, serve_pool, durable, delta, compaction, ttl, tables_using
FROM system.storage_policies;

Disks and pools:

SELECT name, pool, path, role, used_bytes, max_bytes, parts, node_id, type
FROM system.disks;

Parts (immutable segments):

SELECT part_id, relation, shard_id, rows, size_bytes,
       min_key, max_key, disk, created_at_ms
FROM system.parts
WHERE relation = 'trades';

Tiering & movement

Tiered policies move parts between pools by age, size, or pressure. The specifics (pool definitions, movement rules, TTL expressions) are policy configuration.