Oracle 23ai introduces True Cache — a new read-only, in-memory cache tier that sits between your application and the primary Oracle Database. Think of it as a lightweight, always-consistent cache that speaks SQL.
What True Cache is NOT:
- Not Redis or Memcached (those are key-value stores, not SQL-aware)
- Not Oracle In-Memory Column Store (that’s inside the primary database)
- Not a full read replica (it doesn’t write redo logs for all changes)
What True Cache IS:
A lightweight Oracle instance that maintains a hot subset of the primary database in memory, synchronized via redo log shipping. Applications connect to True Cache and run full SQL queries — with full Oracle SQL semantics — but reads are served from memory without hitting primary storage.
Architecture:
App ──► True Cache (in-memory, read-only SQL)
│
│ redo sync
▼
Primary Database (read-write)Key characteristics:
- Consistent reads: True Cache uses the same SCN (System Change Number) mechanism as the primary. Reads are always consistent with the primary database.
- Automatic sync: Changes in the primary are shipped via redo and applied to True Cache — similar to Active Data Guard, but lighter weight.
- No storage footprint: True Cache doesn’t persist data to disk. It’s purely an in-memory acceleration layer.
- Transparent to applications: Applications connect using standard Oracle drivers; no code changes needed.
Best use case: Read-heavy OLTP workloads where the working set fits in memory and read latency matters — product catalogs, session data, reference lookups.
True Cache is Oracle’s answer to the “put everything in Redis” trend, with the critical advantage of full SQL consistency guarantees.