Paper Title: NVMesh: A Cluster Application For NVMe Drives

Paper Abstract: While it is relatively easy and affordable nowadays to add NVMe to a system, turning the individual NVMe drives into something that enables seamless scale-out and is convenient for network access from a cluster is often considered a difficult task, especially without loosing a lot of the NVMe performance advantages. Excelero's NVMesh is a software-defined storage solution specially designed around the advantages of NVMe to solve this problem. NVMesh enables shared NVMe across any network and supports any local or distributed file system. It includes an intelligent management layer that abstracts underlying hardware with CPU offload, creates logical volumes with redundancy, and provides centralized management and monitoring. Customers benefit from the performance of local flash with the convenience of centralized storage while avoiding proprietary hardware lock-in and reducing the overall storage TCO. NVMesh is deployed by a wide range of customers, including cloud and large scale web service companies, finance, data analytics, machine learning, media & entertainment and HPC environments.

Paper Author: Tom Leyden, VP Marketing, Excelero

Author Bio: Sven Breuner decided early in his career to focus on design and development of performance-critical parallel and distributed applications. He joined the Fraunhofer Center for High Performance Computing in 2005, right after receiving his degree in computer science. At Fraunhofer, he developed strong knowledge in storage systems, Linux kernel development, and fast interconnects. This was the basis for his design of the parallel cluster file system BeeGFS. He subsequently lead the BeeGFS development team for several years and in 2014 became the CEO of ThinkParQ, the company that managed the world wide adoption of BeeGFS, including some of the largest and fastest supercomputers in the world. In 2019, Sven joined Excelero as Field CTO to be at the forefront of the paradigm shift to the new generation of NVMe-based storage systems.