65pt intel clear pro presentation titleitoc.sjtu.edu.cn/wp-content/uploads/2019/06/intel-dcpmm.pdf2...
TRANSCRIPT
Chen, Jiang
NOTICES AND DISCLAIMERSIntel technologies’ features and benefits depend on system configuration and may require enabled hardware, software or service activation. Performance varies depending on system configuration.
No product or component can be absolutely secure.
Tests document performance of components on a particular test, in specific systems. Differences in hardware, software, or configuration will affect actual performance. For more complete information about performance and benchmark results, visit http://www.intel.com/benchmarks .
Software and workloads used in performance tests may have been optimized for performance only on Intel microprocessors. Performance tests, such as SYSmark and MobileMark, are measured using specific computer systems, components, software, operations and functions. Any change to any of those factors may cause the results to vary. You should consult other information and performance tests to assist you in fully evaluating your contemplated purchases, including the performance of that product when combined with other products. For more complete information visit http://www.intel.com/benchmarks .
Intel® Advanced Vector Extensions (Intel® AVX)* provides higher throughput to certain processor operations. Due to varying processor power characteristics, utilizing AVX instructions may cause a) some parts to operate at less than the rated frequency and b) some parts with Intel® Turbo Boost Technology 2.0 to not achieve any or maximum turbo frequencies. Performance varies depending on hardware, software, and system configuration and you can learn more at http://www.intel.com/go/turbo.
Intel's compilers may or may not optimize to the same degree for non-Intel microprocessors for optimizations that are not unique to Intel microprocessors. These optimizations include SSE2, SSE3, and SSSE3 instruction sets and other optimizations. Intel does not guarantee the availability, functionality, or effectiveness of any optimization on microprocessors not manufactured by Intel. Microprocessor-dependent optimizations in this product are intended for use with Intel microprocessors. Certain optimizations not specific to Intel microarchitecture are reserved for Intel microprocessors. Please refer to the applicable product User and Reference Guides for more information regarding the specific instruction sets covered by this notice.
Cost reduction scenarios described are intended as examples of how a given Intel-based product, in the specified circumstances and configurations, may affect future costs and provide cost savings. Circumstances will vary. Intel does not guarantee any costs or cost reduction.
Intel does not control or audit third-party benchmark data or the web sites referenced in this document. You should visit the referenced web site and confirm whether referenced data are accurate.
Intel, the Intel logo, Intel Optane, Intel Xeon and VTune are trademarks of Intel Corporation in the U.S. and/or other countries.
*Other names and brands may be claimed as property of others.
© 2019 Intel Corporation.
2
INTEL® TECHNOLOGY COMPLETES THE HIERARCHY
Storage
Memory
Persistent Memory Improvingmemory capacity
DRAMHOT TIER
HDD / TAPECOLD TIER
Delivering efficient storage
Intel® 3D Nand SSD
ImprovingSSD performance
MEMORY AND STORAGE HIERARCHY
Storage
Memory
Persistent Memory Improvingmemory capacity
DRAMHOT TIER
HDD / TAPECOLD TIER
Delivering efficient storage
Intel® 3D Nand SSD
ImprovingSSD performance
10s GB~100ns
10s TB~100usecs
10s TB~10 msecs
100s GB< 1usec
1s TB<10usecs
INTRODUCING
INTEL® OPTANE™ DC PERSISTENT MEMORY
FAST MEMORY SIZE AND DATA PERSISTENCEOF STORAGEEnhance data insights by
Redefining the Memory and Storage Hierarchy
Supported on future
Intel® Xeon® Scalable ProcessorsPlatinum and Gold SKUs
5
INTEL® OPTANE™ DC PERSISTENT MEMORYTHE FAST PERFORMANCE OF MEMORY WITH THE AVAILABILITY AND CAPACITY OF STORAGE
NEW
Future Intel® Xeon® Scalable Processor
INTEL® OPTANE™ DC SSDS INTEL® 3D NAND & SATA SAS & SATADRAM
Smaller CapacityFaster Performance
Larger CapacitySlower Performance
Memory/Storage Type
CHARACTERISTICS
Processing
IT’S MEMORY. IT’S STORAGE. IT’S BOTH. FLEXIBLE AND SCALABLE TO ACCELERATE YOUR WORKLOAD’S DEMANDS AND DATA INSIGHTS
I/O CONTROLLER
APP / WORKLOAD
FILE SYSTEM & DRIVER
USER/APPLICATION
Kernel/OS
6
Data persistence
MORE VIRTUAL MACHINES
INCREASED MEMORY PER VM
EFFICIENT INFRASTRUCTUREWITH COST PARITY
EASE OF ADOPTION
CUSTOMER VALUE AND INITIAL FOCUS WORKLOADS
Hybrid Cloud, IaaS & VirtualizationIncreased VM Density
at the same cost
UpTO 1.2X More VMs
Meeting same slas
Targeting
IMDB & Data ServicesSame Capacity
at significantly lower cost
Fast storage solutionsIncreased storage
at slightly higher cost
UpTO 36TB In a eight
Socket configUpTO 1.2X Data Performance
Per system $ spent
Targeting
TRANSFORM DATA FROM AN EXPENSE TO AN ASSET AS CATALYST FOR GROWTH
IN-MEMORY DATABASES
HIGHER CAPACITY/ LARGER DBS
DATA RELIABILITY/ QUICK RECOVERY
PERFORMANCE OF MEMORY
7
Low Latency Software Path to Persistent Memory
Read(fileptr,offset) /* OS call */Write(fileptr,offest) /* OS call */
ld(address) /* CPU opcode */st(address) /* CPU opcode */
Persistent Memory
Intel® Optane™ DC SSD Intel® Optane™ DC Persistent Memory
Application
Persistent MemoryStorage
Storport
Miniport
File system
Disk Class
Volume Mgr
Part Mgr
Upper filter
Bit locker
File system mini filter
Application
4-10µsfor Linux 1
1. Platform Storage Performance With 3D Xpoint Technology. Frank Hady, Annie Foong, Bryan Veal, Dan Williams, Proceedings of the IEEE. Vol 105, No. 9, Sept 2017 http://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=8003284Towards SSD-Read Enterprise Platforms. Annie Foong, Bryan Veal, Frank Hady. ASMS 2010 – First International Workshop on Accelerating Data Management Systems Using Modern Processor and Storage Architecture. http://www.vldb2010.org/proceedings/files/vldb_2010_workshop/ADMS_2010/adms10-foong.pdf September 2010
DAX Mode
LOW LATENCY SYSTEM ACCESS TO PERSISTENT MEMORY
1 Source: Intel-tested: Average read latency measured at queue depth 1 during 4k random write workload. Measured using FIO 3.1. comparing Intel Reference platform with Optane™ SSD DC P4800X 375GB and Intel® SSD DC P4600 1.6TB compared to SSDs commercially available as of July 1, 2018. Performance results are based on testing as of July 24, 2018 and may not reflect all publicly available security updates. See configuration disclosure for details. No product or component can be absolutely secure.2 App Direct Mode , NeonCity, LBG B1 chipset , CLX B0 28 Core (QDF QQYZ), Memory Conf 192GB DDR4 (per socket) DDR 2666 MT/s, Intel® Optane™ DC Persistent Memory 128GB, BIOS 561.D09, BKC version WW48.5 BKC, Linux OS 4.18.8-100.fc27, Spectre/Meltdown Patched (1,2,3, 3a)
0
25
50
75
100
Idle Average Random Read Latency1
Storage With NAND SSD
Storage with Intel® Optane™ DC SSD
Hardware Latency
Software Latency
Memory SUbsystem with Intel® Optane™
DC Persistent memory
StorageIdle Avg. is About
10µ sfor 4kB
Memory SuBSystemIdle Avg. is About~100ns to ~350ns
for 64B2
Lo
we
r is
be
tte
r
SYSTEM LEVEL PERFORMANCE – HIERARCHY LEVELS
Higher is better
Lo
we
r is
be
tte
r
Intel® Optane™ SSD DC P4800X
Measurement notes:- For SSDs Random
4KB Accesses over entire SSD, Read latency measured per 4kB access
- For Intel® Optane™ DC Persistent Memory Random 256B accesses Over entire module 256B random accesses w/ read latency measured per 64B access
Performance results are based on testing as of dates shown in configuration and may not reflect all publicly available security updates. See configuration disclosure for details. No product or component can be absolutely secure. Software and workloads used in performance tests may have been optimized for performance only on Intel
microprocessors. Performance tests, such as SYSmark and MobileMark, are measured using specific computer systems, components, software, operations and functions. Any change to any of those factors may cause the results to vary. You should consult other information and performance tests to assist you in fully evaluating your contemplated purchases,
including the performance of that product when combined with other products. For more complete information visit www.intel.com/benchmarks. Configurations: see pg 28
OPERATING MODES
BUILT-IN FLEXIBILITY TO USE BOTH MODES SIMULTANEOUSLY
MEMORY MODEPLATFORM/OS/APP ACCESS TO
High SPEED, HIGH CAPACITY MEMORY
“APP DIRECT” MODEAPP/WORKLOAD DIRECT ACCESS TO
HIGH SPEED, HIGH CAPACITY STORAGE
Persistent and much higher data capacity
High availability/less downtime
Significantly faster storage
High capacityTargeting >1.2X More VMs1
Affordable capacity128GB, 256GB and 512GB Modules
Ease of adoptionNo code changes required
11
INTEL® OPTANE™ DC PERSISTENT MEMORY TARGET WORKLOADS WITH APPLICATION SUPPORT FOR LAUNCH
Target List as of 04/02/2019
storage infrastructures databases AI / analytics HPC COMMSIMDB,
Large/Persistent Caching
SAP HANA*MSFT SQL*OracleDB*
Oracle ExadataAerospike*
Redis RocksDB
VM, Container, Application
DensityVMWare vSphere*
MSFT HyperV*KVM*Redis*
Memcached
RDMA/ReplicationOracle Exadata*
Faster StorageMSFT S2D*
Real TimeAnalytics
SAS*
Machine Learning Analytics
Databricks by Spark*
Large MemoryTech Computing
TBD
Content Delivery Network (CDN)
REVENUE-FOCUSED + HIGHEST AEP AFFINITY
APP DIRECT MODE
MEMORY MODE
*Other names and brands may be claimed as the property of others.
12
PERSISTENT MEMORY FOR SEAMLESS CONTENT DELIVERY
DEVELOPER RESOURCES FOR INTEL® OPTANE™ DC PERSISTENT MEMORYFind the PMDK(Persistent Memory Development Kit) at http://pmem.io/pmdk/
Getting Started
Intel IDZ persistent memory-https://software.intel.com/en-us/persistent-memory
Entry into overall architecture -http://pmem.io/2014/08/27/crawl-walk-run.html
Emulate persistent memory -http://pmem.io/2016/02/22/pm-emulation.html
Persistent Memory Programming Video Series -https://software.intel.com/en-us/persistent-memory/get-started/series
Linux Resources
Linux* Community Pmem Wiki -https://nvdimm.wiki.kernel.org/
Pmem enabling in SUSE Linux Enterprise 12 SP2 -https://www.suse.com/communities/blog/nvdimm-enabling-suse-linux-enterprise-12-service-pack-2/
Windows* Resources
Using Byte-Addressable Storage in Windows Server 2016 -https://channel9.msdn.com/Events/Build/2016/P470
Accelerating SQL Server 2016 using Pmem- https://channel9.msdn.com/Shows/Data-Exposed/SQL-Server-2016-and-Windows-Server-2016-SCM--FAST
Other Resources
SNIA Persistent Memory Summit 2018 -https://www.snia.org/pm-summit
Intel manageability tools for Pmem -https://01.org/ixpdimm-sw/
13
UNDER EMBARGO UNTIL 2 APRIL 2019 AT TIME (TO BE DETERMINED)© COPYRIGHT 2019. INTEL CORPORATION.
Customer Pain Points
SAVE MOREHELP
IMPROVE TCO
DRAM-ONLY ALTERNATIVE
DO MORECONSOLIDATE WORKLOADS
INCREASE MEMORY SIZE
GO FASTERADD HIGH
SPEED STORAGE
BREAK THE IO BOTTLENECKS
USE INTEL ® OPTANE™ DC PERSISTENT MEMORY TO…
Storage is too slow
Workloadperformance
Scale up is expensive
DDR4 is too expensive
Operational inefficiencies
Not enough capacity
Performance and Persistence
Infrastructure Consolidation
Memory Price Savings
A HIERARCHY OF INTEL MEMORY TECHNOLOGIES FOR SYSTEM LEVEL APPLICATION INNOVATION
→ Control structures, write log
→ Hot data cache
→ Full large data set storage
MORE VIRTUAL MACHINES
IN-MEMORY DATABASES
DATA PERFORMANCE FOR MEMORY-FOCUSED WORKLOADS
HIGHER RELIABILITY, FASTER RECOVERY. PERFORMANCE OF MEMORY. PERSISTENCE OF STORAGE.
Intel® Optane™ DC persistent memory delivers new capabilities and levels of performance across deployment methods for in-memory databases and virtualization.
INCREASED MEMORY PER VM
EFFICIENT INFRASTRUCTUREWITH COST PARITY
EASE OF ADOPTION
HIGHER CAPACITY/ LARGER DBS
DATA RELIABILITY/QUICK RECOVERY
PERFORMANCE OF MEMORY
16
SUPER-FAST STORAGE
A NEW PARADIGM FOR DATA STORAGE
HIGHEST BANDWIDTH, LOWEST LATENCIES, PERSISTENT AVAILABILITY
Intel® Optane™ DC persistent memory introduces entirely new storage solutionsat levels of performance historically reserved for high-end, memory-only solutions.
CACHING LAYERS
CONSISTENT LOW LATENCY
FASTER PROCESSING AND HIGHER UTILIZATION
HIGHER ENDURANCE AND GREATER BANDWIDTH
CONSISTENT LOW LATENCY
LARGER DATASETS AND FASTER SUPPORT
HIGHER ENDURANCE AND GREATER BANDWIDTH
17
DCPMM OPERATING SYSTEMS (OS) & VIRTUAL MACHINE MONITOR (VMM)
1 The listing of an OS in the table indicates only that it is being used by Intel for DCPMM validation and does not imply Intel has completed DCPMM validation with the OS, or this hardware has full support from the supplier. Our validation goal is not to certify or validate the software.2 Intel is not validating or certifying OS for the DCPMM. Customers are responsible for certification of OS for their platforms. They may validate additional OS and/or platform features that are not validated by Intel and claim support on their platforms.3 Please check with the OS suppliers for the latest and future OS support for DCPMM.
Software ProductGeneral Availability &
Full DCPMM Feature Support 1, 2, 3 Additional Information
Windows Windows Server 2019 Preview Builds Available in Ecosystem Engagement Access Program (EEAP) and Windows Insider Program – please contact your Microsoft Partner Rep for moreinformation.
Latest Windows Server Updates Windows and Hyper-V Persistent Memory Architecture Overview
Windows Windows 10 Pro for Workstations
VMware VMware ESXi– TTM with CLX/DCPMM For more information please contact your VMware Rep. VMware ESX Persistent Memory Architecture Overview
Red Hat Future RHEL 7 update Tech Preview RHEL 7.51,2,3
Configuring and Using Persistent Memory in RHEL 7.3
SUSE SLE15SLE-12-SP4
NVDIMM Enabling in SUSE Linux Enterprise 12, Service Pack 2 NVDIMM Enabling – Part 2 by Intel
Canonical Ubuntu 18.04 LTSUbuntu 18.10
KVM Linux KVM Intel encourages use of latest released Linux Kernel for DCPMM enablement
18
LEARN MORE AT
Supported on future
Intel® Xeon® Scalable ProcessorsPlatinum and Gold SKUs
intel.com/optanedcpersistentmemory
19