Large scale caching systems are making a transition from DRAM to SSDs for cost/power trade-offs and that brings out interesting challenges to both software and hardware. At Facebook, CacheLib is a widely deployed general purpose caching engine, enabling this transition through hybrid caching. In this talk, we introduce hybrid cache, highlight the challenges of hybrid caches at scale and the various techniques Facebook adopts to tackle these challenges.

Boot Devices are a critical component of servers used at scale in the Data Center. The needs and use cases for boot drives in the Data Center are very different than client HDD and Flash based boot drives in laptop applications. This talk will describe the differences between client boot drive use cases and hyperscale use cases. This talk will describe hyperscale use-cases for boot drives, unique needs Hyper-Scalars have for flash based boot drives and hyperscale challenges when deploying flash based boot drives at-scale.

As you know, the Storage Performance Development Kit (SPDK) provides a set of tools and libraries for writing high performance, scalable, user-mode storage applications. Kalray’s MPPA® manycore architecture proposes a unique 80-cores system.

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.

The EDSFF E3 form factors bring more options and commonality that enterprises and hyperscalers desire compared to traditional U.2, but also bring new challenges. With power envelopes increasing in many cases, it is difficult to fit more components on board to maximize capabilities. With a lack of E3 enclosures on the market, there many uncertainties like enclosure geometries and airflow profiles, and the uncertainty of total system power delivery capability.

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.

Big changes lie ahead for the world of data processing in many areas: hardware structure, data management, processing algorithms, and even computing types (analog vs. digital.) How must the technical community prepare for these changes? Join noted analyst Jim Handy and IEEE President Tom Coughlin as they look at future computing systems in which storage will appear in new and interesting places.

/*-->*/

Now that NVMe™ has ratified TP4159 PCIe® Infrastructure for Live Migration, it is time to show the final details of how NVMe controllers in direct attached PCIe SSDs can be migrated from one server to the next for use cases such as workload balancing and maintenance. In this presentation I will detail the new commands, new events, new queues, updates to existing commands, and vendor specific options. This presentation will cover how a host utilizes the new protocol to migrate a controller.