Nano-node: Intelligent Hard Drives in Large Storage Infrastructures
summary
Presents putting a CPU next to each HDD/SSD (using commodity NVMe and local agents) so storage logic—data placement, erasure coding and metadata—runs on the device rather than in large servers. This approach reduces east–west network traffic, speeds rebuilds, limits failure domain to a single device, and yields measurable efficiency gains and simpler scale-out behavior for next-generation workloads.
AI-written summary · about
related talks
09mThe DNA of Next-generation Data C
52mHow ZNS SSD's Are Changing the Landscape
26mLeveraging Moore's Law to Optimize Database Performance
25mPerformance Analysis of NVMe SSDs and their Implication on Real World Databases
31mAI, ML, and the Ceph Advantage: Scalable Storage for Smarter Workflows
13mLiquid-State Drive: A Case for DNA Block Device for Enormous Data
27mPreview: Storage
20mMaximizing the Benefit of Flash Storage for Aerospike