AEP Query Service limits each org to 10 concurrent interactive (ad-hoc) query sessions. A session is defined as a single interactive connection to Query Service — the AEP UI query editor, a PSQL CLI connection, and a BI tool connection each consume one session. Sessions running batch queries do not count against the ad-hoc limit.
Why this matters at scale. Organizations with multiple data teams (marketing analytics, data science, revenue ops) sharing a single AEP org can exhaust the session limit during peak hours. Symptom: "Session Limit Reached" errors appear without warning, blocking analysts mid-query. Unlike the timeout constraint (constraint.aep-adhoc-query-timeout-10min), which can be worked around via query decomposition (pattern.micro-batching-workaround), session exhaustion requires either purchasing additional capacity (Ad Hoc Query Users add-on) or scheduling query windows across teams.
Interaction with composable architecture. In the AEP-as-edge-node pattern, data scientists prefer running large queries in the CDW (Snowflake, BigQuery, Databricks) precisely because CDW session concurrency scales with compute credits — not with a fixed org-level cap. This constraint reinforces the composable-side argument for organizations where query concurrency is a bottleneck.