2. Intel and SAP Collaboration
2
Marketing
and Sales
Platform
Engineering
ArchitecturePlanningPath Finding
Intel® Xeon® processors are
tuned for SAP* workloads.
Intel and SAP have worked
together on the SAP
HANA* platform since
2009.
Intel Xeon processors are
the reference architecture
of the SAP HANA
platform.
The Intel Xeon processor
E7 family is certified by
SAP for use with the
SAP HANA 2 platform.
1 HP, Intel, SAP, SUSE, and VMware. “Virtualizing Enterprise SAP® Software Deployments.” June 2011. suse.com/docrep/documents/o4r7to9d2y/virtualizing_enterprise_SAP_software_deployments.pdf.
2
3. Biggest Data Center Platform Advancement in
This Decade
3
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.
2 Intel estimate, as compared to a representative four-year-old system. For details, see Legal Notices and Disclaimers slides.
Great Performance, Fast
Memory, More
Interconnect/Accelerator
Engines, Advanced
Reliability
Efficient Performance
at Low Power
Best Performance,
Hardware-Enhanced
Security, Best
Business Agility
Entry Performance
3
Intel® Xeon® Processor Scalable Family
4. SAP HANA* Application Memory Support:
8-Socket Intel® Xeon® Processor–Based Systems
4
3 For OLAP, SAP HANA* optimized hardware configurations for SAP HANA SPS 12*. 6 TB compared to 4 TB when running SAP HANA SPS 12 on an
Intel® Xeon® Platinum processor compared to an Intel Xeon processor E7 v4. For OLTP, SAP HANA optimized hardware configurations for SAP
HANA SPS 12. 12 TB compared to 8 TB when running SAP HANA SPS 12 on an Intel Xeon Platinum processor compared to an Intel Xeon processor
E7 v4.
4 SAP. “Find Certified Appliances.” October 2017. https://global.sap.com/community/ebook/2014-09-02-hana-hardware/enEN/appliances.html.
For configuration details, see Legal Notices and Disclaimers slides.
Memory support for online analytical
processing (OLAP)
6x more Memory support for online transaction
processing (OLTP)
3x more
1 TBIntel Xeon
processor E7
Intel Xeon
processor E7 v2
Intel Xeon
processor E7 v3
Intel Xeon
processor E7 v43
Intel Xeon processor
Scalable family4
2 TB
4 TB
4 TB
6 TB
Intel Xeon
processor E7
Intel Xeon
processor E7 v2
Intel Xeon
processor E7 v3
Intel Xeon
processor E7 v43
Intel Xeon processor
Scalable family4
4 TB
6 TB
6 TB
8 TB
12 TB
5. The Intel Vision for the SAP HANA* Platform
Support for any deployment: from sandbox to data center or
cloud
128 GB up to 30 TB64 GB up to 1.5 TB16 GB up to 32 GB
Intel® Xeon® Gold
processor family
Intel Xeon Platinum
processor family
Intel® Core™ i7
processor family
7. Intel Persistent Memory
A true game changer for in-memory computing
– Key business drivers for bringing Intel persistent memory
to in-memory computing:
• As data volumes keep growing, in-memory databases like
the SAP HANA* platform require larger capacities
• DRAM scaling has slowed significantly and will offer
lower capacities
Opportunities:
• Increased scalability
• Significant cost savings
• Improved recovery times
Challenges:
• Higher (than DRAM) latency can impact
performance
• As a new technology, standards are still evolving
8. Intel Persistent Memory
A better option: affordable, large, non-volatile storage
Density
Selectors allow dense packing
and individual access to bits
Scalable
Memory layers can be
stacked in a 3D manner
Persistent
Data/Memory
Instant recovery for
main-memory databases
Total Cost of
Ownership (TCO)
Lower cost per GB
compared to DDR45
5 Technology claims are based on comparisons of latency, density, and write cycling metrics amongst memory technologies recorded on
published specifications of in-market memory products against internal Intel® specifications. Results have been estimated or simulated
using internal analysis or architecture simulation or modeling, and provided to you for informational purposes. Any differences in your
system hardware, software, or configuration may affect your actual performance.
9. Intel Persistent Memory
The future of in-memory database computing
– New, innovative type of memory:
• Persistent, with near-DDR4
DRAM performance
• High capacity: 6 TB per
two-socket system
• More affordable than standard DRAM
Coming at the end of 2018:
• Accommodates a standard DDR4 slot
• Requires a next-generation Intel® Xeon®
Gold processor or Intel Xeon Platinum
processor (codename “Cascade Lake”)
• SAP HANA 2.0 SPS 03* release is designed
to support Intel persistent memory
10. If Storage Devices Were Sprinters
In the time it takes a hard
drive to sprint the length
of a basketball court …
… NAND could finish
a marathon …
... and Intel persistent memory
could circle the globe!6
94 feet 26.2 miles 24,902 miles
6 Intel. “The Wait Is Over! 3D XPoint™ Technology.” 2015. intelsalestraining.com/infographics/memory/3DXPointc.pdf.
11. Intel Persistent Memory Overview
3D XPoint™ technology–based memory module for the data center
• 128 GB, 256 GB, 512 GB
DIMM Capacity
• 2,400 megatransfers per
second (MT/sec)8
Speed
• 3 TB per CPU
Platform
Capacity
11
Non-Volatile
Memory PoolDDR4
DRAM
Flexible, Usage-Specific
Partitions
Memory
mode
Larger memory at
lower cost
App direct mode
Low-latency persistent memory,
persistent data for rapid recovery
Storage over app
direct
Fast direct-attached storage (DAS)
App Direct Storage
Intel persistent memory
8 Dell. “Knowledge Base: Memory information for the 13G
PowerEdge Servers - R630, T630 and R730/XD.” November
2016.
dell.com/support/article/us/en/19/sln294489/memory-
information-for-the-13g-poweredge-servers-r630-t630-
and-r730-xd?lang=en.
12. Increase Your SAP HANA* Data Capacity at
a Lower Cost with Intel Persistent Memory
Persistence with Performance
Fast Restart Times
Affordable Capacity
Intel Persistent Memory for SAP HANA®
GREATER
CAPACITY
OPTIMIZED
TCO
13. DRAM Memory Alternative: Accommodate
More SAP HANA* “Hot” Data In-Memory
Opportunities for higher capacity
Move more data from your storage-area Network (SAN) to SAP HANA in-
memory
3 TB 13.5 TB
15
TB
3 TB DDR4 RAM
(64 GB DIMMs)
DDR4 RAM
1.5 TB (64 GB DIMMs)
Intel persistent memory
12 TB (512 GB DIMMs)
DDR4 RAM
3 TB (128 GB DIMMs)
Intel persistent
memory
12 TB (512 GB DIMMs)
or
small bi
g
14. DRAM Memory Alternative: Accommodate
More SAP HANA* “Hot” Data In-Memory
Opportunities for higher capacity
Move more data from your storage-area Network (SAN) to SAP HANA in-
memory
3 TB 13.5 TB
15
TB
3 TB DDR4 RAM
(64 GB DIMMs)
DDR4 RAM
1.5 TB (64 GB DIMMs)
Intel persistent memory
12 TB (512 GB DIMMs)
DDR4 RAM
3 TB (128 GB DIMMs)
Intel persistent
memory
12 TB (512 GB DIMMs)
or
small bi
g
15. Intel Persistent Memory Accelerates Recovery
Times for the SAP HANA* Platform9
While a backup of the master data is maintained on the
SAN
9 When operating with persistency enabled.
Main
Delta
Temp
SAP
HAN
A
DRAM
Main
1
Delta
1
Temp
Backup
If SAP
Hana
Fails
DRAM
Intel Persistent
Memory
Intel Persistent
Memory
Data
Volume
Log
Volume
SAN
(STORE
S
BACKUP
OF
MASTER
DATA)
• On restart: MAIN is already in Intel
persistent memory, so there is no
need to load data from a SAN
• On hardware failure: a backup server
loads data from the data volume
16. Performance Analysis
Using Intel persistent memory technology for the SAP HANA* database’s
main store
10 Placeholder until test results come in. Source: SAP HANA Adoption of Non-Volatile Memory. VLDB 2017.
– Promising initial results from a prototype
(using hardware emulation):10
• Significant improvements in the restart time
• More than 100x improvement measured
• Acceptable performance impact of higher (than
DRAM) latencies, resulting in slightly lower
performance
• Measured (simulated) performance degradation
was within the expected range for most
workloads
17. SAP HANA and Intel Persistent Memory
– “We are very excited about the benefits Intel DIMMs based on 3D XPoint
non-volatile memory technology will provide for SAP‘s customers, and plan
to make full use of its unique capabilities using App Direct mode.
– With very large capacity and high-performance, servers equipped with Intel
DIMMs will provide a great memory platform for landscape consolidation,
data temperature management, and fast data growth at a significantly
reduced memory cost.
– Additionally, memory persistence on the Intel DIMM will help to enable
near-instant restart times after a power event. SAP and Intel will work jointly
with our open ecosystem to provide these benefits for our SAP HANA
customers
at the introduction of Intel DIMMs.”
Daniel Schneiss
Global Head of Products & Innovation
SAP HANA Platform & Database Products & Innovation, SAP
18. Process More Data Faster to Drive
Innovation
With the SAP HANA* platform and Intel persistent memory
Persistence with Performance
Fast Restart Times
Affordable Capacity
19. Legal Notices and Disclaimers
– 1 HP, Intel, SAP, SUSE, and VMware. “Virtualizing Enterprise SAP® Software Deployments.” June 2011.
suse.com/docrep/documents/o4r7to9d2y/virtualizing_enterprise_SAP_software_deployments.pdf.
– 2 Intel estimate, as compared to a representative four-year-old system. Configurations:
– Up to 1.3x higher virtualized throughput and more VMs/server compared to Intel® Xeon® processor E5 v4 performance estimate based on
virtualization-infrastructure consolidation workload. Baseline: one-node, 2 x Intel Xeon processor E5-2699 v4 with 512 GB total memory on
VMware ESXi 6.0 Update 1* using guest OS Red Hat* Enterprise Linux* (RHEL*) 6, 64-bit operating system, score: 1,034 at 58 VMs compared to
estimates based on Intel internal testing on one-node, 2 x Intel Xeon processor Scalable family–based system.
– Up to 2.7x higher virtualized throughput and more virtual machines (VMs)/server compared to Intel Xeon processor E5 v2 performance
estimate based on virtualization-infrastructure consolidation workload. Baseline: one-node, 2 x Intel Xeon processor E5-2697 v2 (12 cores, 2.7
GHz) with 512 GB total memory on VMware ESXi 6.0 GA using guest OS RHEL 6.4, glassfish3.1.2.2*, postgresql9.2*, and Intel® Solid State Drive
(SSD) Data Center (DC) S3710 Series and Intel SSD DC S3700 Series drives, score: 520.5 at 32 VMs compared to estimates based on Intel
internal testing on one-node, 2 x Intel Xeon processor Scalable family–based system.
– Up to 3.9x higher virtualized throughput and more VMs/server compared to Intel Xeon processor E5 performance estimate based on
virtualization-infrastructure consolidation workload. Baseline: one-node, 2 x Intel Xeon processor E5-2690 with 256 GB total memory on
VMware ESXi 6.0 GA using guest OS RHEL 6.4, glassfish3.1.2.2, postgresql9.2, score: 377.6 at 21 VMs compared to estimates based on Intel
internal testing on one-node, 2 x Intel Xeon processor Scalable family–based system.
– 3 For OLAP, SAP HANA* optimized hardware configurations for SAP HANA SPS 12*. 6 TB compared to 4 TB when running SAP HANA SPS
12 on an Intel® Xeon® Platinum processor compared to an Intel Xeon processor E7 v4. For OLTP, SAP HANA optimized hardware configurations
for SAP HANA SPS 12. 12 TB compared to 8 TB when running SAP HANA SPS 12 on an Intel Xeon Platinum processor compared to an Intel Xeon
processor E7 v4.
19
20. Legal Notices and Disclaimers
– 4 SAP. “Find Certified Appliances.” October 2017. https://global.sap.com/community/ebook/2014-09-02-hana-
hardware/enEN/appliances.html.
– Up to 6x greater system memory supported compared to available solutions from four years ago (representing the currently installed data
center base). SAP has certified its SAP HANA 2 platform for OLAP workloads to support up to 3 TB of memory per system for the upcoming
Intel® Xeon® processor Scalable family for a 4-socket configuration (or 6 TB for an 8-socket configuration). Systems available four years ago
(representing the typical data center installed base infrastructure) could only support 0.5 TB (or 1 TB in an 8-socket configuration). For
comparative purposes, SAP certifies support for up to 2 TB of memory for the current Intel Xeon processor E7 v4 family in a 4-socket
configuration, so upcoming Intel Xeon processor Scalable family–based systems are certified to support up to 50 percent greater system
memory that the generation they replace.
– Up to 3x greater system memory supported compared to available solutions from four years ago (representing the currently installed data
center base). For OLTP workloads, SAP has certified its SAP HANA 2 platform to support up to 6 TB of memory per system for the Intel Xeon
processor Scalable family for a 4-socket configuration (or 12 TB for an 8-socket configuration). Systems available four years ago (representing
the typical data center installed base infrastructure) could only support 2 TB (or 4 TB in an 8-socket configuration).
– 5 Technology claims are based on comparisons of latency, density, and write cycling metrics amongst memory technologies recorded
on published specifications of in-market memory products against internal Intel® specifications. Results have been estimated or simulated
using internal analysis or architecture simulation or modeling, and provided to you for informational purposes. Any differences in your system
hardware, software, or configuration may affect your actual performance.
– 6 Intel. “The Wait Is Over! 3D XPoint™ Technology.” 2015. intelsalestraining.com/infographics/memory/3DXPointc.pdf.
– 7 Low latencies: Because data will be transferred across the memory bus, latencies will be orders of magnitude lower when compared
to transferring data across Peripheral Component Interconnect Express* (PCIe*) or an input/output (I/O) bus to NAND/HDDs. Benchmarks and
proof points are forthcoming.
– 8 Dell. “Knowledge Base: Memory information for the 13G PowerEdge Servers - R630, T630 and R730/XD.” November 2016.
dell.com/support/article/us/en/19/sln294489/memory-information-for-the-13g-poweredge-servers-r630-t630-and-r730-xd?lang=en.
– 9 When operating with persistency enabled.
20
21. Legal Notices and Disclaimers
– 10 Placeholder until test results come in. Source: SAP HANA Adoption of Non-Volatile Memory. VLDB 2017.
– Benchmark and estimated results were obtained prior to implementation of recent software patches and firmware updates intended to address
exploits referred to as "Spectre" and "Meltdown." Implementation of these updates may make these results inapplicable to your device or
system.
– The benchmark results reported above may need to be revised as additional testing is conducted. The results depend on the specific platform
configurations and workloads utilized in the testing, and may not be applicable to any particular user’s components, computer system, or
workloads. The results are not necessarily representative of other benchmarks and other benchmark results may show greater or lesser impact
from mitigations.
– 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 information
go to 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 intel.com/go/turbo.
– Results have been estimated or simulated using internal Intel analysis or architecture simulation or modeling, and provided to you for
informational purposes. Any differences in your system hardware, software or configuration may affect your actual performance.
– 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.
21