OCP Storage Project Workshop @ Seagate: January 24th, 2019

Description (Please note that the audio begins around the 3 minute mark of the video):
OCP presents another opportunity to meet face to face at the OCP Storage Workshop. Hosted by OCP Gold Member Seagate, this will give you a wonderful platform on collaboration on open storage solutions. Among the topics of discussion will be the Fast Fail initiative (a new API initiative from the Cloud HDD subgroup), Archival storage, and more.

AGENDA
Intro and Welcome - Jason Adrian & Jorge Campello

HDD Connector considerations for PCIe/NVMe - Seagate

There are a number of connector standards being utilized today for PCIe connected SSDs. As we look to the opportunity of utilizing the PCIe/NVMe standards for HDDs, selecting the right connector to achieve easy adoption & optimal cost points for the industry can presents some challenges. Seagate would like to review some analysis and options for PCIe/NVMe connectors with the community as a means to foster ideation for further collaboration.
We would also like to propose the formation of an OCP working group, similar to the Tape Durability Modeling effort, to continue discussions and make progress toward defining some standards for physical and logical connection design to enable NVMe for HDD technology.

34 Drive 2 OU JBOD - Gooselake - Inspur

Introduction to Inspur's 34 drive, 2 OU JBOD

Ethernet & RDMA tape drive connectivity - Spectra Logic (Matt Ninesling)

Starting with a SCSI interface tape has progressed to fibre channel and SAS. With the release of TS1155 tape now has the option of a high speed Ethernet interface with RoCE as well. Providing high throughput, little additional overhead, and low latency, RoCE is an interface that makes sense for today’s modern data center that utilizes a high speed Ethernet backbone. It is a newer technology though that has not been widely adopted. Does RoCE make sense for tape and if so how can we make it easier to use?

Break

Fast Fail Read (Cloud HDD subgroup) - Cloud HDD Subgroup (Lawrence Ying)

Sharing the concept, specification, and use case of Fast Fail Reads for HDDs

Fast Fail Read - T10 updates - Western Digital (Ralph Weber)

Combining the OCP Fast Fail Read spec with the “Disks for Data Centers” White Paper from the Fast 2016 Conference, a proposal for additions to various SCSI standards has been developed, reviewed, and refined. This presentation will be a guide through the several layers of the proposal culminating with a thorough description of the key functional changes. The presentation will be based on the latest proposal revision, i including changes requested by the January 2019 T10 CAP (Commands, Architecture, and Protocols) working group.

Key Metrics to Monitor for Tape Media Health - Ed Childers

High Performance HDDs - Seagate

Dual actuator demo