Samsung has introduced the CXL Memory Module – Box (CMM-B), a breakthrough in orchestrating the vast array of memory options for storage systems. This leading-edge technology has been developed to fit seamlessly into Supermicro’s Plug and Play Rack Scale Solutions.
Samsung has introduced the CXL Memory Module – Box (CMM-B), a breakthrough in orchestrating the vast array of memory options for storage systems. This leading-edge technology has been developed to fit seamlessly into Supermicro’s Plug and Play Rack Scale Solutions. It promises an accelerated pathway to productivity while effectively reducing the Total Cost of Ownership (TCO).
CMM-B is an advanced memory pooling appliance tailored for rack computing, utilizing Compute Express Link (CXL) technology. It supports eight E3.S CXL Memory Modules – DRAM (CMM-D) within a rack appliance, providing versatile memory resource allocation.
Image provided by Jianping Jiang, XConn Technologies Holdings Inc.
Samsung CXL Memory Module – Box
The CMM-B is pivotal in creating the industry’s premier Rack-Level Memory Solution, designed for a highly scalable and composable disaggregated memory infrastructure. It facilitates flexible memory resource allocation by harnessing CMM-B alongside memory pooling, allowing memory capacity and bandwidth to scale without hitches. The architecture avoids memory stranding and supports on-demand memory capacity and bandwidth expansion. The solution integrates a rack-scale memory bank (CMM-B), application hosts, Samsung’s Cognos management console, and a top-of-rack (ToR) switch, all co-engineered by Samsung and Supermicro, to shorten the operational timeline for Rack Scale Solutions.
This technology creates a memory architecture that provides the following:
- Disaggregated Memory Allocation: The architecture enables shared access over a network by presenting remote memory as a pool, decoupling compute and memory resources for independent allocation.
- Composable Memory Orchestration: Facilitates unifying and sharing memory pools across multiple host systems.
- Extensible Memory Expansion: Provides the capacity to merge memory resources, forming larger pools that can be dynamically allocated as needed.
The Rack-Scale solution provides the infrastructure for a complete memory server solution. All the hardware and software in one rack space allows fast, easy-to-use memory allocation. But that is only one of many benefits of CMM-B. Rack-scale management Integration of Samsung Cognos Management Console with a Rack-Scale Management solution (for example, Supermicro’s SuperCloud Composer) enables CMM-B to be composable.
Image provided by Jianping Jiang, XConn Technologies Holdings Inc.
Rack-scale composable memory bank features and benefits include:
- Composable Rack Scale Management: The integration with Samsung Cognos Management Console and Supermicro’s management solutions renders CMM-B highly adaptable.
- Advanced Memory Pooling: Compatible with CXL1.1 and CXL 2.0 protocols, CMM-B supports a sophisticated pooling mechanism manageable via REST API through the SCMC console.
- Scalable Device Configuration: CMM-B’s expansion capabilities allow for an increase in both memory capacity and bandwidth, supporting large-scale CMM-D integrations.
- Energy Efficiency: Implementing the SoC switch chip under CXL protocols promotes power-saving operations, thus optimizing TCO.
- Dynamic Memory Allocation: The Samsung Cognos Management Console (SCMC) facilitates autonomous memory allocation independent of server constraints and ensures memory is allocated where it is most required.
- Memory Allocation Analytics: SCMC provides comprehensive visualization tools for monitoring memory allocation, aiding in the assessment and optimization of memory configuration.
- User-Friendly Software: With state-of-the-art tools, CMM-B simplifies the utilization of the CXL memory stack, enhancing the management and expansion of memory resources in the era of AI-driven computing.
After years of development, the CXL Consortium and its member companies have achieved a significant milestone, providing a solution to address the memory wall issue in modern AI computing. Samsung, Supermicro, and XCONN Technologies are driving rack-scale computing platforms.
Engage with StorageReview
Newsletter | YouTube | Podcast iTunes/Spotify | Instagram | Twitter | TikTok | RSS Feed
