SNIA Developer Conference September 15-17, 2025 | Santa Clara, CA
Sergei Platonov is the Vice President of Strategy at Xinnor, a technology company specializing in storage solutions. Sergei's journey in the tech industry began in 2010 when he graduated from the university, earning a degree in Computer Science. With a strong passion for storage technologies, he joined SDS start-up in 2009. Throughout his career, Sergei has been instrumental in the development of Storage Area Networks (SAN) and has focused on the research and implementation of cutting-edge technologies such as Flash, NVMe (Non-Volatile Memory Express), CXL (Compute Express Link), and storage performance improvement techniques. His expertise in these areas has enabled Xinnor to create highly efficient and reliable storage solutions that cater to the evolving needs of businesses. Sergei's contributions to the industry go beyond development work. He is a recognized thought leader and has delivered more than 30 presentations at conferences, sharing his insights and knowledge on storage technologies. Additionally, he has authored numerous articles, shedding light on the advancements in Storage Class Memory and its impact on storage performance.
The performance of network file protocols is a critical factor in the efficiency of the AI and Machine Learning pipeline. This presentation provides a detailed comparative analysis of the two leading protocols, Server Message Block (SMB) and Network File System (NFS), specifically for demanding AI workloads. We evaluate the advanced capabilities of both protocols, comparing SMB3 with SMB Direct and Multichannel against NFS with RDMA and multistream TCP configurations. The industry-standard MLPerf Storage benchmark is used to simulate realistic AI data access patterns, providing a robust foundation for our comparison. The core of this research focuses on quantifying the performance differences and identifying the operational and configuration overhead associated with each technology.
In modern cloud environments, there is an increasing demand to provide high-performance block volumes within virtual machines. Many existing solutions come with several drawbacks: low performance and high latency, heavy system resource consumption, and sometimes the necessity to install additional client software. We aim to discuss various approaches that enable volume passthrough within virtual machines: vhost and vhost-user-blk, vfio-usr, NVMe/TCP, NVMe/RDMA. We will share the problems we identified and how we resolved them. The presentation will feature extensive performance data and a detailed bottleneck analysis. As a result, we will demonstrate how to achieve performance levels of tens and hundreds of Gbps and several million IOPS within a virtual machine using only a few CPU cores. Special attention will be given to the use of DPUs.
The performance of network file protocols is a critical factor in the efficiency of the AI and Machine Learning pipeline. This presentation provides a detailed comparative analysis of the two leading protocols, Server Message Block (SMB) and Network File System (NFS), specifically for demanding AI workloads.
In this presentation, I will delve into the world of NVMe RAID and present a comprehensive analysis and comparison of various NVMe RAID engines. Through comparative testing and performance analysis using state-of-the-art performance analysis tools, I have identified bottlenecks in these engines when handling different workloads. During the presentation, I will provide valuable insights and comparisons, showcasing significant figures that highlight the performance differences among the tested NVMe RAID solutions.
