Episode 6. Exadata. Day 6 of your 10 day sprint.

Welcome. Exadata moves filtering into the storage cells so the database scans less. Offload is the whole game, and eligibility decides it.

Say this sentence: Offload only what the cell can filter, functions and binds decide.

Concepts. Smart scan does predicate filtering, column projection, and decompression inside cells. Watch cell smart table scan versus index scan efficiency to prove offload happened. Blockers: functions wrapping indexed columns, bind type mismatch, chained rows, unsupported encryption paths. Partitioning by the query predicate plus local versus global index choice keeps payments queries pruned.

Scenario. A daily report stops offloading after a code change. Understand: which query regressed and when. Prepare: cell offload efficiency before and after, plus the changed predicate. Identify: a new function on the filter column killing smart scan; rewrite the predicate first. Support: deploy the rewrite, confirm offload returns, add an offload-efficiency check to deploy gates.

Say this bridge once. Say: Partitioning and cell offload are my scale tools on Exadata, and the same access-pattern-first thinking designs shard keys and sort keys anywhere.

Your Day 6 lab: take one reporting query, list its offload blockers, and sketch a partition scheme by its most common predicate.
Teaser: Tomorrow, Episode 7, HA DR and sev handling. Customer impact first, root cause second.

Second scenario. Flash cache hit rate collapses on mixed workload. Understand: which workload arrived and cache hit trend. Prepare: IORM plan plus celldisk balance. Identify: a new batch scan evicting OLTP cache; IORM throttle the batch first. Support: set the IORM plan, verify hit rate recovery, document the workload mix rule.

Pause and say the offload sentence again. Functions and binds decide. Twice.

Close with one STAR seed: an Exadata scale win with a scan-time number. Done.