Oracle Exadata: Architecture and internals technical deep dive

on YouTube ↗22m

" summary

Exadata moves predicates to storage and returns only needed columns, implementing Row Filtering and Column Projection, verified by AWR and SQL Monitor. Storage Index keeps per-1MB min/max and bloom filters to skip IOs (example: 588GB scanned, 252GB saved; another case 173GB → 49MB, ~3,000x reduction). Flash cache stores columnar formats for vectorized processing (664GB processed, 445GB avoided). Persistent-memory RDMA yields sub-19µs single-block reads and >2 million IO/s to a storage server at near 0% CPU.

AI-written summary · about

" chapters (5)

1. 0:00Start
2. 4:13Cloud Storage Bottleneck
3. 5:06Traditional Scan
4. 10:57Advances in Storage Index
5. 15:37Superior OLTP Architecture - Memory Cache Acceleration

" related talks

• 19mSpeed up transparent performance with Exadata: What, when, how, and whyKOTHANDA UMAMAGESWARAN · Oracle DatabaseWorld 2022
• 32mExadata: New features and functionalityKOTHANDA UMAMAGESWARAN, SUBBAN RAGHUNATHAN · Oracle DatabaseWorld 2021
• 24mFuture of Data and AIJuan Loaiza · Oracle DatabaseWorld: Multicloud AI Edition 2025
• 27mDatabase In-Memory: Real-time analytics, superchargedANDY RIVENES, Shasank Chavan · Oracle DatabaseWorld 2021
• 19mMission-critical databases made simpleMaria Colgan, Ashish Ray · Oracle DatabaseWorld 2022
• 13mOptimizing Oracle Database for MulticloudTammy Bednar · Oracle DatabaseWorld: Multicloud AI Edition 2025
• 18mNext Gen Data Architecture for MulticloudAshish Ray · Oracle DatabaseWorld: Multicloud AI Edition 2025
• 24mOracle's vision for simplifying data-driven apps and analyticsJuan Loaiza · Oracle DatabaseWorld 2022