User Guide > User Guide > Using Transient Tables
Was this helpful?
Using Transient Tables
Introduction to Transient Tables
Transient tables are global objects that bridge the gap between session temporary tables and permanent tables and have the following features:
They are visible across multiple concurrent sessions with the appropriate privileges.
They are registered in the iirelation, iiattribute, and iioutstanding system catalogs.
They persist until they are explicitly dropped with DROP TABLE, or until the Analytics Engine (X100) server restarts.
To preserve a transient table across one X100 server restart, use CALL X100(transient_snapshot) followed by COMMIT. The snapshot is one-shot — it is consumed by the first X100 restart that replays it; a subsequent restart with no fresh snapshot discards the data.
They are designed to provide high-performance, cost-efficient storage for staging massive datasets without the overhead of backup, recovery, or high availability (HA) replication.
They cannot share the same name as a permanent table and consume physical storage.
The iioutstanding Catalog tracks transient tables, so the DBMS can clean up stale entries on database open after the X100 server restarts without a fresh snapshot. Cleanup also removes dependent system-generated procedures, functions, and iidbdepends rows for the dropped tables.
Last modified date: 09/11/2026