Timezone isn't accessible, please provide the correct parameters
eventFeedUrl=http://realintelligence.com/customers/expos/00Do0000000aAt2/FMS_xmlcreator/a0J1J00001H0ji2_specific-event-list.xml
trackCategory=Session
eventID=a0J1J00001H0ji2
timezone=
duration=PTH
, NaNth
3:40-4:45 PM
ENST-102A-1: Enterprise Storage Design (Enterprise Storage Track)
Paper Title: Using Storage Class Memory to Accelerate All Flash Storage - Lessons Learned

Paper Abstract: With 10 years of building hybrid storage arrays, using NAND flash memory as a caching layer in an HDD-based storage array, the authors organization embarked upon building an array using SCM as a caching layer on top of a NAND flash storage layer. Building functioning versions of this next generation hybrid was straightforward. Unlocking the performance potential of this combination required non-trivial engineering. Challenges encountered included the requirement to support the full performance of a storage controller using IOPS delivered from a single slot, the requirement to fully utilize the entire performance potential of a single, SCM-based SSD, and supporting response-time requirements an order of magnitude lower than previous systems. All of these issues had to be addressed in the context of a storage system that is committed to always-on inline dedupe and compression, including for blocks stored in the caching layer. This paper will provide a brief overview of the system architecture, enumerate the performance challenges inherent in this class of design, and discuss the design trade-offs that we made. We will conclude with some measured performance results.

Paper Author: Stephen Daniel, Distinguished Technologist, HPE / SPC

Author Bio: Stephen Daniel has spent over 30 years working the design and implementation of high performance commercial computing systems, the last 5 years at HPE Nimble Storage. For the past 10 years Mr Daniel has been a representative on the Storage Performance Council (SPC) and currently serves as the Chair of the SPC Steering Committee. During 13+ years at NetApp, Mr. Daniel worked on storage issues related to database performance, features, service levels, and technologies. During 12 years at Data General Mr. Daniel worked on various portions of Data General’s UNIX kernel with an emphasis on system performance and file system design. Mr. Daniel has a Masters degree in Computer Science from Duke University.