Computational storage is a new paradigm in computing where data processing is moved closer to the storage device to enhance performance and reduce data transfer bottlenecks. In this presentation, we showcase how data integrity checks on CSDs with a software toolkit and a dedicated hardware to process the host payload can significantly improve storage performance and reduce data transfer overhead. This is a cross-industry effort that includes the system, the system software, and the CSD representing a complete end-to-end solution.

Azure Container Storage is a container native, microservices-based storage service providing unified volume management across different cloud storage backends, enabling consistency and portability. Azure Elastic SAN, which is one such backend, is a purpose-built, highly available block storage solution designed to scale to millions of IOPS at low ms latency. In this talk, we will deep dive into how Azure Container Storage uses Azure Elastic SAN to provide elastic bursting of stateful pods for large scale data processing at low cost.

The Open Programmable Infrastructure (or OPI) is an open-source effort within the Linux Foundation to develop a standard API for utilizing SmartNICs, DPUs and IPUs, and other coprocessors or processing elements. It will allow users to provision and orchestrate all devices in the same way, thus allowing them to handle many different devices, implement new devices, and change or replace devices without learning a new command structure. It will also allow manufacturers to create a standard API, deliver new or upgraded devices faster, and benefit from a large ecosystem.

The long-term retention and backup requirements of many organizations continue to grow as their data estate grows. The long-term preservation market provides an opportunity for a higher durability service to store copies of high value digital assets in as many places, with as many types of media as possible, to eliminate the chance of data loss. The ability to store large volumes of such “long-term preservation” data, largely depends on enabling storage technologies that are lowest cost than any storage technology that exists today, such that cost to store data is near zero.

A prominent trend in disaggregated storage is the use of Non-Volatile Memory Express over Fabric (NVME-oF) and in particular, NVMe over TCP to connect storage devices over a network. But there is no straightforward to provide this storage to Virtual machines and containers. The hypervisor will still need to emulate virtio-blk or virtio-scsi kind of emulated interfaces to expose this storage which involves usage of hypervisor processor cycles.

Modern data centers invariably face performance challenges due to the rising volume of datasets and complexity of deep learning workloads. Sizeable research and development has taken place to understand AI/ML workloads. These workloads are computationally intensive, but also require vast amounts of data to train models and draw inferences. The impact of storage on AI/ML pipelines therefore merits additional study.

The conflicting needs of datacenters managers hosting 3rd party and internal data securely and storage vendors needing to have a stream of vendor unique fleet telemetry for monitoring and debug has historically not found a scalable solution. This paper describes how a new proposal driven from the OCP Storage Workgroup facilitating standardized telemetry to be securely shared with storage vendors enabling vendor deep learning failure analysis and debug. The approach even allows vendor specific telemetry in a standard way vs current solutions such as NVMe SMART.

Should SSDs supporting power states higher than the maximum TDP dissipation supportable in a system? Many industry standards for drive form factors are targeting =25W, but will Gen6 SSDs be viable in a x4 configuration or will these form factors be abandoned? What is proposed is a framework currently supported in NVMe and OCP's Datacenter NVMe SSD Specification of allowing enhanced latencies in cases where there is thermal margin above the maximum TDP of the form factor using either host orchestrated NVMe power state management or device orchestrated Host Controlled Thermal Management.

Abstract: Fibre Channel extends its renowned compatibility and reliability with a new speed, 128GFC. This talk will discuss the newly completed 128GFC specification as well as uses of Fibre Channel in storage disaggregation and machine learning. The speakers, Rupin Mohan and Craig Carlson, have decades of experience in the architecture and standards definition of Fibre Channel and storage systems.

Tail latencies in networking tend to worry us all, whether we implement distributed storage and compute or whether we connect systems-of-systems in automotive or factory automation, for example. Same goes for the computational burden of processing networking protocols. One of the foundations of reliable networking is TCP, the Transmission Control Protocol which was introduced half a century ago. Today, TCP is ubiquitous: In the datacenter, in mobile communication, the Internet and in (embedded) systems-of-systems.