The Power of Multi-Model: KV, Document, Graph
summary
Aerospike runs a hybrid memory architecture that keeps only an index in RAM and treats NVMe/SSD as addressable memory, demonstrated on a five‑node cluster holding 800 million objects (≈6 TB on disk) while using ~95–100 GB RAM. The workload sustains about 300k ops/sec (≈200k reads + 100k writes) with P99 ≈ 1 ms and costs roughly $7,926/month on AWS including licensing.
AI-written summary · about
chapters (6)
- 0:00Introduction and welcome
- 1:10Graph performance demonstration
- 3:20Multi-model database overview
- 5:54Hybrid memory architecture explained
- 7:59Demo results and cost efficiency
- 9:29Use cases and future outlook
related talks
28mAerospike's Journey Through Innovations, Sustainability, and Roadmap
24mCloud Workloads At Petabyte Scale
26mLeveraging Moore's Law to Optimize Database Performance
26mLeveraging Moore's Law to Optimize Database Performance
15mAWS On Air Voice of the Customer ft. Aerospike, Inc.
45mAerospike + Persistent Memory = Voila!
22mThe Art of the Possible
15mUnderstanding the Power of Graph Databases