CXL is an open standard industry-supported cache-coherent interconnect for processors, memory expansion, and accelerators. Essentially, CXL technology maintains memory coherency between the CPU memory space and memory on attached devices. This enables resource sharing (or pooling) for higher performance, reduces software stack complexity, and lowers overall system cost.  It uses existing PCIe bus. More info on CXL is here https://computeexpresslink.org/about-cxl/

CXL enables memory disaggregation. It enables use cases where memory can be shared across multiple hosts using CXL switches and other interesting mechanisms.

In Micron, We are working on FamFS (Fabric Attached Memory Filesystem), which enables the shared memory use case where multiple hosts can access same memory and run applications on them. Famfs has a kernel component and a user space component.

FamFS once created on a node, can be used (RO) in other nodes to which the underlying CXL card is physically connected through a CXL switch.

This talk aims to provide an idea on FamFS and its use cases.

I have around 18+ years of experience in Software design and development, a lot of that in open source development. I have previously presented in SDC - India, SDC - US, FMS and other conferences.

I am an active contributor/Maintainer for this project https://github.com/cxl-micron-reskit/famfs. The main developer/designer for this project is John Groves. I joined Micron recently and contributing to it very recently.