Ultra-Reliable Low-Latency Solutions for Industry 4.0 Predictive Maintenance
summary
Benchmarks show Aerospike on HPE servers with Intel Optane PMEM sustaining 14.87 million TPS in a read-heavy test; a payments customer cuts cluster nodes by two‑thirds, meets 99.99% uptime and 10–20 ms processing windows. A ScaleFlux computational drive yields a 3x CPU improvement and ~80% latency reduction versus NVMe, handling >160 threads. Use cases include predictive maintenance for wind turbines and railways fed into an in‑RAM ML pipeline (Parquet, Python plugin, Vertica) and a NiFi–Kafka–Flink NiFi–Kafka–Flink pipeline for real‑time COVID‑19 ingestion.
AI-written summary · about
chapters (22)
- 0:00Intro
- 1:23Aerospike
- 4:00Data Science Lab
- 7:05Commercial Drones
- 8:07Inspection Use Cases
- 9:10Rail Inspections
- 9:42Wind Turbine Blades
- 10:28Industry V4 Use Cases
- 13:55Semantic Web
- 14:49GraphDB
- 15:47Aerospike vs GraphDB
- 17:00AI Driven Pipeline
- 19:17Team
- 20:03HPE PMEM
- 22:02RealTime Pipelines
- 23:41ScaleFlex
- 24:22What is ScaleFlex
- 25:33Observations
- 28:23COVID19 Data
- 29:02RealTime Pipeline
- 32:40Future Work
- 33:28Outro
related talks
32mFuture-Proofing your Architecture - AI at Any Scale with HPE & Aerospike
08mBuilding a Real-time Data Future
11mThe Power of Multi-Model: KV, Document, Graph
19mSpark vs. Accelerated Compute: A Balanced Look at Data Processing Strategies
27mThe convergence of machine learning and high performance computing
20mBuilding realtime efficient queries in an IOT platform
41mBeyond Uptime: Revolutionizing Fintech Reliability
24mLarge Scale On-Demand Low-Latency Near Real-Time Predictions