Speed up transparent performance with Exadata: What, when, how, and why
summary
Explains how Exadata sends queries to storage so storage does row filtering and column projection, cutting bytes returned by up to 3000X in examples (173 GB scanned returning 49 MB) and reporting 588 GB scanned with 252 GB saved by storage index. Describes automatic in‑storage columnar conversion into flash cache, RDMA to persistent memory serving single‑block reads under 19 microseconds and redo commits averaging 16.79 microseconds, and >440 SQL operators offloaded.
AI-written summary · about
chapters (16)
- 0:00Intro
- 0:43Exadata Vision Extreme Performance and Availability.Lowest Cost. Available Everywhere
- 2:30Storage Networking Background
- 3:00Traditional Monolithic Storage Bottleneck
- 3:20Cloud Storage Bottleneck
- 4:02Traditional Scan
- 5:50Smart Scan Filtering and Projection in AWR Report
- 8:57Advances in Storage Index
- 9:55Storage Index in AWR Report
- 10:15In Memory Analytics Capacity in Shared Storage
- 11:09Columnar Cache in AWR Report
- 11:49Smart Scan Performance across Storage Servers
- 12:28Superior OLTP Architecture - Persistent Memory Cache
- 13:20Superior OLTP Architecture - Persistent Memory Commit
- 14:15Persistent Memory Log in AWR Report
- 16:40Exadata Benefits
related talks
22mOracle Exadata: Architecture and internals technical deep dive
32mExadata: New features and functionality
14mMeeting database management business challenges with Oracle Exadata
27mDatabase In-Memory: Real-time analytics, supercharged
30mExadata warehouse: simplify capacity planning and forecasting for Exadata
24mFuture of Data and AI
13mOptimizing Oracle Database for Multicloud
29mOracle TimesTen: extreme performance for latency-critical applications