By: Anil Godbole, Senior Marketing Manager for Intel’s Xeon Product Planning and Marketing Group
Modern data centers are under constant pressure to keep pace with rapidly expanding datasets and increasingly complex workloads. Traditional, fixed memory configurations often create inefficiencies and bottlenecks that limit performance.
This demonstration shows how CXL memory pooling enables systems to dynamically share and allocate memory across processors, boosting capacity and workload scalability. The setup integrates four Intel Granite Rapids-AP servers with a CXL switch connected to 22 Micron CZ122 memory expansion devices, collectively forming a 5.6 TB shared memory pool accessible by all participating servers.
Each server can dynamically augment its local DRAM capacity by allocating memory from the shared pool, enabling real-time responsiveness to demanding analytical workloads. During complex query execution, when local memory resources are exhausted, the server interacts with the CXL memory orchestrator, which provisions additional memory segments while maintaining strict isolation and access control policies.
This architecture facilitates:
- Dynamic memory scaling across heterogeneous compute nodes
- Efficient resource sharing and reclamation
- Improved memory utilization and workload performance
By breaking traditional memory boundaries, CXL memory pooling delivers a flexible and scalable solution tailored for high-performance, real-time analytics in enterprise and cloud data centers.
Experience the Demo at the CXL Pavilion
Visit Intel’s kiosk at the CXL Pavilion (Booth #817) to explore how CXL memory pooling powers high-performance analytical databases. Together with the CXL community, Intel is driving the next generation of high-performance, composable infrastructure to unlock flexibility, scalability, and efficiency for data-intensive computing.