In-Memory Databases: Business Intelligence & Analytics

AI Inferencing: Recommendation System   |   AI Inferencing: Chatbot Services   |   HPC: Computer Aided Engineering

2x In-Memory Database Capacity, 1.5x More Transactions and Query Times Cut in Half

Limited Memory Bandwidth for Multi-Processor Big Data Analytics
Without CXL

Challenge:

  • Multi-socket systems are popularly used for expanding memory
  • Memory bottlenecks limit connections, transactions, and queries
  • Disk I/O operations can greatly slow down online transactions and analytics
  • Untuned memory tiering solutions may lead to unpredictable performance

Expanded Memory Bandwidth for Multi-Processor Big Data Analytics
With CXL

Solution:

  • CXL enables power efficient memory expansion with fewer CPUs
  • Doubling memory capacity enables more concurrent database users
  • Doubling memory bandwidth helps reduce latency and improves CPU utilization
  • Leo offers plug-and-play support with CXL hardware interleaving

Videos