Memory Workloads

This guide demonstrates operational usage of MonkDB memory objects for session/context storage patterns.

1) Create base incident table

CREATE TABLE incidents (
  incident_id TEXT,
  summary TEXT,
  created_at TIMESTAMP
);

INSERT INTO incidents (incident_id, summary, created_at) VALUES
  ('inc-1', 'payment failure in eu', CURRENT_TIMESTAMP),
  ('inc-2', 'card auth timeout', CURRENT_TIMESTAMP);

2) Create memory object

CREATE MEMORY session_store
WITH (strict = true, entry_ttl_ms = 3600000, max_entries = 1000);

3) Inspect memory metadata

SELECT memory_name, options['strict'], options['entry_ttl_ms'], options['max_entries']
FROM memory.memories
WHERE memory_name = 'session_store';

4) Insert key-value entries

INSERT INTO memory.memory_entries (memory_name, entry_key, entry_value)
VALUES
  ('session_store', 'sess-42:intent', 'payment_failure'),
  ('session_store', 'sess-42:priority', 'urgent');

5) Read entries

SELECT memory_name, entry_key, entry_value
FROM memory.memory_entries
WHERE memory_name = 'session_store'
ORDER BY entry_key;

6) Upsert with conflict handling

INSERT INTO memory.memory_entries (memory_name, entry_key, entry_value)
VALUES ('session_store', 'sess-42:intent', 'payment_failure_v2')
ON CONFLICT (memory_name, entry_key) DO UPDATE
SET entry_value = excluded.entry_value;

7) Compact and monitor

ALTER MEMORY session_store COMPACT;

SELECT last_compacted_at IS NOT NULL,
       last_compaction_pruned_entries,
       total_compaction_pruned_entries
FROM memory.memories
WHERE memory_name = 'session_store';

SELECT strict, entry_ttl_ms, max_entries, active_entries, expired_entries, capacity_used_percent
FROM memory.memory_status
WHERE memory_name = 'session_store';

8) Cleanup

DROP MEMORY session_store;