Make IT Simple Make Business Agile Huawei HPC overview Dr. Kunbin HONG Director of IT Solution WEU Enterprise
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Make IT Simple Make Business Agile
Huawei HPC overview
Dr. Kunbin HONGDirector of IT SolutionWEU Enterprise
2
Huawei at a Glance
3
228Ranking in
Fortune Global 500(Jul 2015)
3.5 B 31 ~76,000 ~170,000
EmployeesEnd usersJoint innovation
centersR&D
Employees
170+
Countries
A Global Leader of ICT Solutions
3
Position as ICT Solution Provider
Consumer Business
A brand loved by consumers
Carrier Business
Customers' best strategic partner
Enterprise Business
Enterprises' best innovation partner
ICT Product and Solution
ProviderInformation Generation Information Transmission Information Processing and Storage
Data center infrastructure
Big Data analytic
platform
Cloud services
Smartphones
MBB & home
devices
Wearable devices
Vehicle telematics
Mobile network
Fixed network
Telecom software
and core network
Enterprise network
M2M connection
management platform
Serving global carriersServing hundreds of
millions of consumersFocusing on valued industries
Professional services
4
Huawei Performance 2015 (internal estimate)
35.4
32.4
27.6
46.5
39.5
2010 2011 2012 2013 2014
*
Revenue
(USD in billion)
2015
61.8
* Internal estimate, not audited yet,
Carrier
networksEnterprise Consumer
~16% ~35% ~68%
32+%
y2014
5
Huawei Outlook: 10+% CAGR In the next 5 years
Carrier BusinessTo create more value through
software and services
Well-positioned to achieve
steady growth
Enterprise BusinessPromising addressable market for Huawei
in the enterprise business
Strong competitiveness and growth
potential
Consumer Business Huge market demand
Our product capabilities are
advancing quickly
50+ Billion $By 2020
50+ Billion $By 2020
6
Continual Innovation Investment
Customer requirements driven
Cutting-edge technologies driven
* : As of Dec. 31, 2014
** : KPMG-audited results announced early Q2, 2015
76,000 employees dedicated in R&D
Joined 170+ standard & open source organizations
Holding 180+ key positions in IEEE-SA, ETSI…
38,825 patents granted*
$ 6.6 billion (in 2014 )
$ 29.8 billion (from 2004 to 2014)
USD in million
4831
5074
6592**
2005 2006 2007 2008 2009 2010 2011 2012 2013 20142004 2015
$ 9 Billion
(2015
est.)
7
Joint Innovation with Leading Customers
……
31 Joint Innovation Centers
User insight
Business Insight
Customer
Engagement
+
Innovative solutions
Better Customer
Experience
Cost Reduction
Value
Enhancement
Green ICT
Joint Business
Success
R&D Expertise
Global Practice
Huawei
Engagement
8
Future-oriented technology Innovation
MirrorsysAll-
optical
network
5G mobile
communicationsData Center
3.0
Network
5.0
MirrorSys：Offer next-generation fully immersive user experience
5G mobile communications: continuously enhance MBB and enable Internet of massive things
All-optical network: Break the bottlenecks of transport networks and build limitless transport capacity
Network 5.0: Pervasive mobility, massive IoT, content aware and content-based-routing switching network
Data Center 3.0: Applications affinitive data center( HPC, big data processing and analysis…)
Information generation Information transmission Information processing and storage
9
Nurturing Ecosystem with Openness and To-be-integrated Strategy
4strategic alliances
300solution partners
6,300channel partners
1,200service partners
10
Huawei Services - Enterprise Business Partners
Huawei Enterprise Full Lifecycle Services
PARTNER
Professional
Service
Maintenance
Support
Training &
Certification
CUSTOMER
Enterprise
Networking
UC&C
IT Product
Enterprise
Wireless
CoCare
Integration ServiceL1/L2 Customer
SupportTraining Services
Installation &
CommissioningConsulting Services
Validation
Deployment
Migration
Network Integration
Hardware Support
Software Support
Solution Support
Managed Support
OptimizationNetwork Design
Integrated Design
Consulting
Assessment
Planning
Pro
du
cts
& s
olu
tio
ns
HiC
are
11
Build
Enterprise Full Lifecycle Service
• Consulting
• Assessment
• Application
• ICT Design
• Industry Integrated
Design
• ICT Implement
• Site Services
• Solution Implement
• Migration
• Hi-Care
• Co-Care
• Specialized Services
• Validation
• Optimization
Plan Services Design ServicesImplementation
ServicesOperation Services
OptimizationServices
Plan Run
Training & Certification Services
Unified ICT Service: Accelerates Your Business
12
UC&C
Products
Security Products
Unified ManagementBring Your
Own Device
BYOD
Software Defined
Network SDNCloud
Data Center
ConferencingTelepresence IVSUC Contact Center
Data Center
InfrastructureServer Storage Cloud Computing Networking Power
Enterprise
Network & Wireless
DC Network OTN/MicrowaveGSM-R/LTE
ICT Product and Solution Portfolio
Agile Network
13
SME/Branch
Worldwide
Enterprise
Micro DC
Modular DC
Container DC
Scalability/Reliability
Cloud
Datacenter
ManageOne
DC Management4U 4S
RH5885
ES3000
SSD Card
RH2288
2U 2S
RH1288
1U 2S
OceanStor Dorado2100 G2
OceanStor Dorado5100
OceanStor S2600T
OceanStor 9000
OceanStor18000 V3 Series
RH8100
OceanStor 5300 V3
OceanStor 5600 V3
OceanStor 5800 V3
OceanStorS2200T
E9000
FusionCube
9000
FusionCube
6000
Hyper-Converged
Infrastructure
X6800
FusionInsight
FusionAccess
FusionSphere
FusionStorage
FusionStage
APP
Mission Critical
9016/9032
Overview of Huawei Enterprise IT Products
14
eS
igh
t/ U
20
00
: Un
ified
Ne
two
rk M
an
ag
em
en
t
E2E IP Products & Solutions
The Only Company that Provides E2E IP “Network Products”
Security
Products
NIP2000/5000SVN 2000/5000USG9500 AntiDDoS1000/8000ASG2000 AVE2000 TSM/AnyOffice/iSOC/
UMA/DSM/OMMUSG 2000/5000
DC
security gatewayNGFW UTM Security access
gatewayIDS/IPS AntiDDoS Antivirus gateway
USG6300/6600
WAF
WAF2200/55200
TSMOnline behavior
management
Full Series of
Ethernet
Switches
Terabit core switch Intelligent routing
switchGigabit switch
S9700 S5700/S5710 S3700 S2700S7700 S6700 S1700
Data center switch
CloudEngine12800CloudEngine
5800/6800
Layer 3 100M
switchSMB switchAgile switch
S12700
Layer 2 100M
switch10G aggregation
switch
Access &
Service
Gateways
Service gateway BRAS
ME60 X3/X8/X16
xPON OLTMxU ONT
MA561x/2x(A)/
3x/9x/7x/58xxMA5652/
MA5669MA560x/
MA568x
HG80x0(H)/81xx(F)/82xx
(H)/8x45A/824xT/8240F
Access &
Core Routers
Wi-Fi
NE20E-X6AR1200AR150/160/200 AR2200 AR3200
Access router AR G3 Core router Wi-Fi AP Wi-Fi AC
AC6605
SPU 2.0AP7110SN/DN
AP6010SN/DN
AP6310SNAP6510DN
/AP6610DNNE40E-
X3/X8/X16 NE5000E
Industrial router
AR530
AP2010DN
AP5010SN/DN
AP5030DN
AP5130DN AP7030DENE16EX
WDM
OSN6800OSN1800I/II/V OSN8800 OSN9800 RTN 310/380OSN3500 OSN3580 OSN7500/7500II OSN9560
PTN950/960 PTN910/910-F PTN1900/3900-8/3900
PTN
MSTP/Hybrid MSTP
OSN500/550/580Metro100/1000 OSN1500
Microwave
RTN910 RTN950 RTN980OSN2500
OTN &
Microwave
Products
15
So
lutio
ns
So
ftware
Pro
du
ctsH
ard
ware
Pro
du
cts
S2200T/S2600T 18500/18800 9000
Massive Storage
DeviceManager
(Device Management)
eBackup
(Virtualization Backup)(Unified Management)
All-flash array
2 controllers
500 s latency
Smart Series Hyper Series Info Series
Dorado2100 G2 Dorado5100
(Data Protection Software)(Intelligent Resource Management Software) (Big Data Storage Value-Added Software )
InfoTierInfoEqualizer InfoAllocator
InfoExplorer InfoProtector
3-288 nodes
60 PB capacity
400GB/s throughput
SmartTier SmartMotion SmartQoS
SmartThin SmartPartition SmartVirtualization
HyperCopy HyperSnap HyperClone
HyperReplication
2-16 controllers
3 TB cache
3216 disks
SPC-1: 1,005,893 IOPS
Unified Storage
MediaHPC AnalysisRetrievalDR & Backup Active-ActiveFile Sharing Virtualization Database Acceleration
Converged Storage Solutions Big Data Storage Solutions Flash Acceleration Solution
2 controllers
8 GB cache
276 disks
UDS
Amazon S3
EB-level scalability
2 PB+ capacity per
cabinet
eSight
Video
Surveillance
5300 V3/5500 V3 6800 V35600 V3/5800 V3
2-8 controllers
512 GB cache
750 disks
2-8 controllers
1 TB cache
1250 disks
2-8 controllers
4TB cache
3200 disks
Storage
Integration
Huawei Storage – OceanStor Product Portfolio
16
Intelligence on Demand, Convergence for the Future
Unified Storage V3
5300, 5500, 5600, 5800, 6800Block + File
Mass Data Storage
UDS
Object for Nearline & Offline Market
All-in-one Mass Data Storage
OceanStor 9000
File + Object + Analysis + Archiving
P15P22P1P8
P15P22P1P8
P15P22P1P8
P15P22P1P8
P15P22P1P8
P15P22P1P8
Unified block & file storage
In-depth technical
convergence
Advanced conception
For global cloud data centers:
Decentralized & distributed
architecture & highly efficient
global data sharing
All-in-One storage solution
simplifying data lifecycle
management
Complete HUAWEI OceanStor PortfolioVirtualization, SSD, Distributed Storage, Software Defined Storage
Full flash Storage
Dorado 2100G2 & 5100Block
Full flash storage: best ratio
of performance/price
17
No.4
worldwide HuaweiHP DELL Cisco Inspur FujitsuLenovo
Source: Gartner Quarterly Statistics: Servers, 2Q15 Update
No.4
For 9 consecutive quarters,
Huawei server shipments rank
Global Mainstream Server Vendor
18
2002 2008 20152012 2018
Engineering
Application
solutions
Chips
Architecture
Innovation Makes Computing Simple
Innovation Reliability Openness
Telecom Internet Key industries
DEMT dynamic energy saving,
PCIe hot swapping, high-voltage
DC
Modular server, Project Scorpio
cabinet, PCIe SSDs, FPGA
acceleration
Internet application server (web,
memory, database, etc.)
ASIC mgmt chip, RAID
controller, convergent I/O
chip
Orthogonal backplane,
memory riser hot swapping,
25G network, liquid cooling
32-socket fault-tolerant
server, hyper-converged
infrastructure
SAP HANA appliance,
virtualization appliance, HPC
solution
Black box, high temperatures
(55°C)
ATCA architecture
x86 midrange computer
All industries
SSD controller (new), network
controller (new)
CPU hot swapping, optical
interconnection, efficient data
center network protocol,
resonant antenna
RAS 2.0 structure, fault
database-based PFA, full
decoupling DC 3.0
Deduplication, compression-
featured VDI appliance,
SPARK appliance, RISC-to-IA
midrange computer migration
SSD controller
19
Underlying chip innovation
Stable and reliable foundation
Chips
SSD storage
control
Processor
interconnectionManagement
Network
control
RAID
control
Innovation Reliability Openness
20
Surpassing common servers
High availability
Engineering InnovationQuick fault locating
Black box design
Hot swap
Memory module
HVDC
High-voltage DC
Dynamic energy
saving
DEMT
40ºC
High-temperature
resistance design
High speed
Orthogonal
backplane
Innovation Reliability Openness
21
Energy Certification
Country and Region
Europe Union North America China
Certification CE-HVDC CB-HVDCTUV-HVDC
Energy Star SPECpower 80PLUS UL-HVDC CCC-HVDC CQCEnvironm
entEnergy
Blade Energy
standard
Huawei Servers provides Green products to customer and obtained
most green energy certification in different areas. Including: Energy
Star、80PLUS、CQC、China Environment Energy、HVDC、SPEC etc.
(Few Pattern Rights (DEMT, PID etc.))
Energy
efficiency
22
Fully explored system potential
High reliability, performance,
energy efficiency
Architecture InnovationFirst use
of commercial ATCA
on a large scale
First release
hyper-converged
system
First in China
cabinet server
First use
RAS2.0
Innovation Reliability Openness
23
Customer oriented
Excellent user experience
Application Solution
Innovation
World's fastest SAP
HANA
Virtualization IOPS up 10–36 fold; Database TPM performance up
2.67–9.7 fold
Innovation Reliability Openness
24
Innovation Reliability Openness
Design
Process
Components
Quality is
ReliabilityAlibaba Quality Award
Tencent Best Supplier Award
Failure Rate 15% Lower
than Industry Average
• RAS2.0
• Shock absorption
• Passive backplane
• Advanced heat-proof
design
• Stringent supplier
vetting
• Industrial-grade
components
• FMEA (20 years+)
• Derating design
• Fault warning
• CPU, hard disk, and
memory check tools
• Professional reliability
test
• Comprehensive
authentication
25
Stable running in 45°C Shock absorption design, reduces
the device fault rate
Note: The optimized air vents, high-efficiency fan
modules and PID adjustment of fan speed, support
45°C operating temperature, 10°C higher than that
supported by mainstream servers (10°C-35°C).
The comprehensive optimization design ensures the reliable running of the device, and the fault
rate is the lowest in the industry.
Note: 1) The fan component shock isolation and hard disk tray shock absorption design are designed to avoid HDD-sensitive frequencies and to decrease risks of HDD performance deterioration caused by rotating components.2) Reasonable packaging is designed to
effectively reduce the impact of shocks and to ensure system reliability during the transport.
Carrier-class components
and derating design
Polymer capacitors
X7R capacitors are required in thermal-
sensitive areas
Gold-plated connectors
Note: 1) All use Huawei carrier-class high-reliability
components and authentication process.
2) The derating design of Huawei servers
lowers the electric and temperature stress
on components, delaying the parameter
degradation, prolonging the life cycle, and
increasing the reliability.
Innovation Reliability Openness
Stable and Reliable System Design
26
Leading Reliability
Independent air-duct, improving
system heat dissipation
Quality control throughout the whole
process
The most stringent standards of
reliability testing for components
Passive backplanes avoid single
points of failure
• Front and rear independent
air-duct
• Duct design optimization
• High and low temperature test,
HALT
• 3000 times Power cycle test
• Continuous 30 days stress test
• Supplier Certification
• Component selection and
control
• IPD process from module to
system
• 150 reliability patents
• EMI testing, safety testing
HUAWEI Others
Failure Rate(According to the statistical result from customer T)
0.87% 1.12%Support stable operation at 40°C high temperature, ensuring business continuity
The failure rate is 15% lower than competitors
Ensure Reliability by Advanced Design and Strict Testing
• The passive backplanes contain
no active components and no
resistors, capacitors or ICs,
except PCBs and connectors
Innovation Reliability Openness
27
Load Balancing across disks
Shorter Recovery time of data on faulty disks
Quick Reads and Writes
Adjust the capacity of a LUNs dynamically
Create Storage pools with different type of disks
Available for Huawei Oceanstor Products
Huawei innovation --Declustered Raid (RAID2.0+)
in storage system
28
• Density optimized rack
mount server
• Rack scale system
X6800
• Optimized Hardware for
Hyper-converged
appliance
• Vmware, Openstack,
fusionsphere, etc
• Open
• RAID 2.0
Sca
le u
p
Mission Critical
Applications
Converged Infrastructure
E9000 X6000
X8000
iNIC
Scale outConverged
Multi-Node High Density System
NVMe SSD card
RH5885
RH8100
RH2288
KunLun
16S+ System
for Mainframe/
Mini Downsizing
Server Development planning directions
29
Scale-out: High Density, Energy Saving, and Increased
Deployment Efficiency
X8000
X6000
X6800
Large space
required
High energy
consumption
Slow
deployment
2U, 4 nodes
4U, 8 nodes
44U, 80 nodes
Energy saving
Space saving
Rapid deployment
30
Converged Architecture Meeting Various Workload Demands
Private cloud and hybrid cloud
Unified resource management and
flexible expansion
Enterprise key services
Converged IT infrastructure
Acceleration appliance
PCIe SSDs and distributed storage eliminate
the disk I/O bottleneck
HPC
up to 16.5 TFLOPS per chassis and
speedy interconnection.
In-memory computing
SAP HANA cluster based on E9000
FusionCubeStorage
Computing Network
Management
6 types of compute nodes
11 types of switch modules
Supporting
GE/10 GE/FCoE/FC/IB
Unified management
15 x 2.5-inch disks in a slot
ES3000 PCIe SSD
31
Scale-up: Full Range of Enterprise Key Applications Servers
RH5885H V3
4PRH8100 V3
16P+
The Scale-up x86 Platform
8P
NC chip
Virtualization
Virtualization Performance upgraded 70%
Business Intelligence (BI)
Leading TPC-E performance, best choice for
high-performance key database.
In-memory computing
The world’s fastest SAP HANA appliance
Enterprise Critical
Ensures effective handling of workload with high
reliability and low TCO to deliver high performance
and efficient resource utilization.
High Reliability
High Performance
32
800K IOPS
>12个SATA SSD
Improved>100%
VDI densityReduced 90%
Data analysis time
ES3500P V3
2.0/3.2TB
• Support NVMe, UEFI Bootable
• Hot-swappable, Out-band management
• NVMControl: Best IO performance
• SR-IOV: improve VM IO performance
• Atomic Write: improve DB performance
VDIDatabase Indexing ServerSAN
ES3600P V3
1.2/1.6/3.2TB
BDA
Application Accelerator: ES3000 V3 NVMe PCIe SSD
3.2 GB/s BW
>6个SATA SSD
33
MAXIMIZE EFFICIENCY ACCELERATE WORKLOAD
MAXIMIZE PERFORMANCE FOR
INDIVIDUAL WORKLOAD
Flexible, modular architecture
Multiple innovative form factors
Deep optimization with hardware
acceleration
Super Fat nodes
CONVERGED HPC & BIG DATA
MAXIMIZE HARDWARE ROI
Single HPC cluster and storage
system for both traditional HPC
MPI workload and Hadoop
Innovative big data analytic
appliance with deep hardware
and software optimization for
maximum cost effectiveness
Simplify HPC systems today
MORE COMPUTE LESS SPACE
LOWER POWER CONSUMPTION
End-to-end energy efficient design
HVDC
High Ambient Temperature ~40oC
Direct Liquid Cooling ~ 84% coverage
Tight integration with Huawei data
center infrastructure
SDS
Big Data
SDI
34
Values of Huawei HPC Solution
Optimal
Performance
Energy
Saving
Fast
Delivery
35
Huawei HPC Solution Portfolio
Parallel file systemParallel environment System
deployment
Customized
development
Backup and
restoration
Compiling and development environment
Industry
applications
Service
platforms
CAE/CFD Life sciencesAerospaceAnimation rendering
Meteorological environment Physics & chemistry
GPGPURack server Cabinet storageXeon Phi
Rack storage
Solid-state storage GE switchIB/10GE switchBlade server
Windows Linux
+
Computing Storage Network
Hardware
resources
OS Computing environment
System
environment
CHESSPBS Works Platform JH Scheduler Cloud renderingBright Cluster Manager
Cluster
management
Container data centerModular data center
Infrastructure
36
Optimal Performance : High-density Solution
High-Performance Servers
Compute node
Highlight
42＋TFLOPS Computing
Capability/Chassis
10 Year Network Evolution
100GE,16G/32G FC, IB EDR
6 Type of nodes Housing
Max 64 CPUs
Thin,Fat,IO,Management
E9000Switch
37
20＋TFLOPS Computing
Capability/Chassis
Optimal Performance : Accelerator Solution
High-Performance Servers Highlight
16× ES3000
PCIe cards
8× double slots
GPGPU/Phi
X6800
Storage node
I/O node
Xeon Phi
GPGPU
Compute nodes
38
Optimal Performance : Memory-Intensive Solution
HPC Fat Node: KunLun 9032 with up to 32TB Memory Space Per Single Node
The world’s
First 32P x86Mission Critical Server4S
8S
16S
32S
Dell Lenovo HP Huawei
RAS 2.0Unique CPU & memory
hot-swapCPU
CPUMemory
Memory
• Innovative, 32 interconnected CPUs, industry leading
performance and scalability
• Network Chip designed by Huawei
• Reliable, RAS 2.0 technology, supporting hot-swapping of
CPUs and memory modules
• Open, mainstream x86 ecosystem, 30% lower TCO than
UNIX servers
• Intel® Xeon® E7-4800/E7-8800 v3 series processors
• Up to 24 TB (32 GB DIMMs), 32 TB (64 GB DIMMs)
• Up to 768 DIMMs (12-DIMM memory risers)
• Highest SPECint2006 performance in the market for 32way
servers
39
Distributed NAS Highlight
Optimal Performance : Storage Solution
400GB/s Bandwidth
linear enhancement
100PBThe biggest file system
N+M data protection
A fully symmetrical distributed
architectureOceastor9000
Backend
IB or 10GE
Front-end :
IB or 10GE
GE
management
40
Green Design Highlight
Energy Saving: Liquid Cooling Solution
40％ Energy Saving
45 kW/Cabinet Heat
Dissipation Capability
9 Reliability technologiesLiquid Cooling
41
Modular architecture, 1 week Installation Highlight
Fast Delivery: All-In-Room Solution
70% less delivery period
25% less initial investment
Modular for easy expansion
Management
Server
Storage
Monitoring
Network
Power supply
42
Energy Efficient Server Design
Bottom line of Greener IT
Higher Temperature
Up to 8%↓ WW DC power consumption w/ 5ºC↑
HVDC
Up to 9-15% conversion efficiency
improvement
Free Air Cooling
Up to 20-70% cooling saving
In-Memory Computing
Up to 90+% less power
consumption than HDD-based
SSD Storage
Up to 60+% less power consumption than
HDD-based
Air Containment
Up to 30%↑ in cooling efficiency
Green IT Reduces Energy Bill & CO2 Emission, Extends DC Life, Lowers TCO
Hot Water
Eliminate Chiller Free cooling
Right-sized Power
~20% better power & cooling utilization
End to End Green HPC Design
Hardware Acceleration
~ 2X+ Performance Per Watt
43
Model ES3500P V3 ES3600P V3
Capacity 2TB/3.2TB 1.2TB/1.6TB/3.2TB
Form Factor 2.5“, SFF-8639 connector, PCIe 3.0 x4
NAND Flash 15/16nm MLC
Max Read Througput 3GB/s 3.2GB/s
Sustained Read IOPs
(4Kb)
750k 800k
Max Write Througput 2GB/s 2.1GB/s
Sustained Write IOPs 60k 170k
Endurance 1 DW/D, 5 years 3 DW/D, 5 years
Average Read/Write
Latency
89µs&14µs
MTBF 2 million hours
Power Consumption Idle: 4W, max: 25W
NVMe ES3000 - only 25W under full load
Proprietary Huawei ASIC
44
HPC Joint Innovation with Openness and Collaboration
OS
System
Integration
Hardware
Cluster
Management
Application
CAE Rending Life sciences Climatic Geoscience
45
Huawei HPC World Wide Success Cases
Central Asia
ChinaEurope
North America
South America
University of Nebraska-Lincoln
University of Tennessee, Knoxville
Digital Domain
Macau University of Science and Technology
GlobalFoundries
Kyushu University
Institute of Disaster Prevention
Hebei Environmental Protection Agency
Beijing Data Communication Research
Institute
Beijing Jiaotong University
BeiHang University
Southwest University
Zhejiang University
Capital Medical University
China Electric Power Research Institute
Shanghai Aerospace Power Machinery
Research Institute
ZhongXin Biotechology -HPC
bioinformatics based cloud platform
Sino-Singapore Tianjin Eco-city
Project Phoenix
Beijing Forestry University
Shanghai Institute of Satellite Engineering
Henan Environmental Protection Agency
Zhuozhou Geophysical Bureau
Turkish Academic Network and
Information Center
Yıldız Technical University
Istanbul Technical University
Harran University
Mexico Water Conservancy Bureau
The Ministry of agriculture of Mexico
Universidade Presbiteriana Mackenzie
Chile observatory
Poznan Supercomputing and
Networking Center (PSNC / PCSS)
Interdisciplinary Centre for
Mathematical and Computational
Modeling, University of Warsaw (ICM)
Gdansk University of Technology
(TASK)
Daimler AG
Newcastle University in U.K.
Illumination Mac Guff HPC
St.Petersburg State University HPC
Deltares, Netherland
Ehrenburg Water Research Institute
Rijkswaterstaat
Universidad De Burgos
Southeast Asia
46
CFD Simulation HPC platform for Mercedes-Benz, Germany
In automotive design and development process, more and more dependent on
technology for CAE-aided design, such as automotive exterior design, aerodynamic
simulation, collision simulation.
existing HPC systems can not meet requirement of CAE, urgent need to improve
system performance and meet the strict requirements to energy consumption
Challenges
17 X6800 with 132 computing nodes, to improve HPC system performance;
X6800,4U with 8 XH620 V3, Intel Xeon E5-2600 V3,192GB DDR4, excellent
computational density.
Shared power supply& fan design, innovation duct design and efficient
power ,reduce energy consumption. long-term operation at 40 ℃
Solution
With Start-ccm + test, performance improvement of 31%, significantly increasing the
efficiency of CAE systems;
Reduce energy consumption by 10%;
50% space-saving compared with traditional rack servers , reducing space occupied
in the equipment room .
Customer Benefits
I believe you can do it, Now，You are a manufacturer with
top-level Benchmark.
---- Mercedes-Benz, Germany
As the world‘s top automobile brand, Mercedes-Benz is the world’s
largest luxury car manufacturer, the largest commercial vehicle
manufacturer, the world‘s top 500 ranked No. 17.
47
Huawei Innovate Manufacturing CAE Solution with ESI
IC.IDO Virtual Reality Solution
• Immersive 3D experience in real time and
real scale
• Interactive & eye-catching presentations
facilitate processes
Virtual Performance Solution
• Support multiple domains of industry
• Shorten product development cycles
• Improve simulation performance by 50%
with tuning software & hardware
Huawei Virtual Manufacturing Solution
48
HPC Rendering Platform for Illumination Mac Guff, France
• The animation rendering system must deliver excellent performance to address heavy computing workload and frequent data access.
• Illumination Mac Guff's studio is located in downtown Paris. It is costly to expand the data center area. The equipment room space is limited.
• High stability is critical to servers in the animation rendering system to ensure business continuity.• Since the studio has limited IT O&M personnel, servers must support efficient batch deployment
and management. The server vendor must offer responsive local services.
Challenges
• Combined with CH140 high-density compute nodes, the Huawei E9000 can provide sixty-four 130 W high-performance CPUs in a 12U chassis to deliver high computing density.
• The E9000 integrates the chassis management modules and iBMC management software to support unified software deployment and upgrades for all compute nodes in a chassis. The E9000 also supports stateless computing.
• Huawei ensures high reliability of the E9000 by using stringent component selection, independent ventilation channels, and passive midplane design. The E9000 can stably operate at 40°C (104°F).
Solution
• Enhanced rendering performance: The E9000 supports a maximum floating-point computing performance of 16.5 TFLOPS per chassis, which improves rendering performance by 80%.
• Less equipment footprint: The E9000 can provide 64 CPUs in a 12U chassis to deliver ultra-high density and save equipment room space by 50%. The E9000 can also operate stably at 40°C.
• Simplified deployment and O&M: The time required for a rack to be fully configured with compute nodes has remarkably decreased from over 10 hours to roughly 30 minutes. The E9000 supports automated fault detection and reporting, and configuration migration to improve O&M efficiency.
• Rapid service rollout: Unified management and service provisioning of the E9000 have shortened the TTM by 30%, meeting customer demands for rapid service rollout.
Customer Benefits
"The service platform that supports animation production to
achieve special effects faces many IT challenges, including
system scalability and tiered storage. To respond to these
challenges, Huawei deploys cloud computing, server, and
storage resources to help us produce animation films that are
popular with the audience."
---- Bruno Mahe, Head of Technology, Illumination Mac Guff
49
HPC System for Newcastle University, UK
Customer Benefits
Solution
• Rapid growth of computing data: single node requires 512 GB
memory, overall computing capability > 88 TFLOPS
• I/O read and write bandwidth: > 1 GB/s
• System can be easily expanded
• Rapid deployment: Integrated computing, network, and storage
resources reduces service deployment time to three days.
• Simple management: Unified management minimizes
maintenance workloads.
• Easy expansion: easy expansion to reduce later investment costs
and meet expansion requirements for the next five years.
• Huawei All-in-One HPC solution: 122 E9000 blade server nodes,
S5500T storage device, full-10 GE high speed network, Huawei
cluster management software, with Lustre high-performance,
parallel storage.
• The All-in-One HPC modular solution meets requirements for future
expansion.(So far, completed four phases deployment)
Challenges
Bioinformatics Medicine
R&D
StorageComputing
Network
Huawei HPC solution
"We are impressed by Huawei's HPC solution that integrates
computing, storage, and network resources. Huawei's HPC solution
provides us with an easy-to-expand and cost-effective HPC cluster.
This is the solution we need."
— Jason Bain, Deputy Dean of the IT Infrastructure and
Architecture Department, Newcastle University
Molecular
dynamics
simulation
50
Supercomputer for ICM faculty, University of Warsaw, Poland
• Existing HPC devices can not meet the requirement of performance
and business
• Long time construction period, complicated management
• Cloud evolution in the future, providing commercial service and
internal use
Challenges
Solution
• 240 * CH121v3 blade node (Haswell E5-2697V3CPU)
• 56Gbps FDR Infiniband Network
• Lustre parallel storage
Customer Benefits
• Optimal performance: Rpeak =279.5TFlops
• Energy saving: 2935.85Mflops/Watt
• Efficient management: Intuitive graphical operations and various scheduling
algorithms
Interdisciplinary Centre for Mathematical and Computational Modeling
PL-GRID Plus project funded by EU was established in 2012, and its goal is to build a HPC
platform for Polish and international scientific research and commercial applications within t
hree years
13 application area: Astronomy, HEP, nano-technology, acoustics, life sciences, quantum
chemistry and molecular physics, ecology, energy, bioinformatics, health sciences, materi
als, metallurgy
51
Background
Benefits
Liquid cooling case: Poznan Supercomputing and Networking Center
High Performance Computing centre within PSNC:
provides computing power, disc space and archiving
systems for science, business and public institutions;
appears on the TOP500 list of the most powerful computing
systems in the world.
total computing power 1.4Peta-Flops, TOP80 in TOP500
(2015 Nov, http://top500.org/list/2015/11/)
Innovation – Direct Liquid Cooling
Efficiency – 80% heat removed by liquid
Openness – Support Partner for Customization
52
Alibaba: Hadoop Compression Card for Massive Data
The large data amount leads to the increasing servers Compressing the existing data occupies too many CPU resources and require high CPU
performance The algorithm of high compression rate accompanies with low performance; the
algorithm of high compression performance, low compression ratio
Customer’s Pain Points
Huawei-developed compression card can significantly improve the data storage space utilization, and has small impact on the current program, with characteristics such as transparent usage, lightweight materials, high acceleration performance, and low CPU usage.
Huawei's Solution
The hardware uninstallation GZIP algorithm reduces the CPU usage The CPU usage is lower than 20%, decreasing by 80% The compression performance can reach 400 MB/s, decompression performance 680
MB/s, which is over 10 times the software performance A standard ZLIB enables upper-layer services to use hardware compression cards
transparently The storage space is reduced by more than 50% The power consumption is decreased by more than 80%
Customer Benefits
53
Huawei has incomparable ICT solutions and deep
understanding of Big Data technologies. We were
deeply impressed by the superior quality and quick
response of its servers.
– Matthieu Blumberg, Global Technology Director of Criteo
Criteo : Biggest Hadoop cluster in Europe
Criteo, headquartered in Paris, France, is the second largest personalized retargeting company that provides
personalized online display advertisements to consumers in the world. Criteo is dedicated to offering accurate
advertising services for advertisers by leveraging advanced data analysis algorithms.
Criteo plans to build the largest Hadoop Big Data computing cluster in Europe. A Big Data analysis platform
consisting of 5000 servers will be built to accelerate data processing and improve accuracy of predictive
advertising.
Challenges
Huawei provided 600+ RH2288H V3 servers to build the Big Data computing cluster in the first phase.
The Huawei RH2288H V3 server uses the latest E5-2600 v3 series processors, supports large-capacity disks
and virtualization, and provides the industry-leading SPEC performance. It is the ideal choice for the Hadoop
platform.
Huawei servers use innovative heat dissipation design and highly efficient power supplies, achieving optimal
density and energy efficiency.
Solutions
The Hadoop platform built with Huawei RH2288H V3 servers provides outstanding performance and improves
Big Data analysis performance by 30%.
Equipped with 6 TB of hard disks, the RH2288H V3 servers meet data storage requirements of the Hadoop
platform, reducing the number of external storage devices.
The innovative energy-saving and heat dissipation design reduces the platform power consumption by 10%.
Customer Benefits
54
Joint Lab Incubation
• High Temperature Server（Baidu）• Research how to build up a server can support high temp(40°C) in a free air cooling DC
• Neural Network Acceleration based on GPGPU an FPGA（Baidu）
• Heterogeneous computing（Alibaba）• ARM 64bit Enablement and Optimization
• SSD Customization（Alibaba）• Various endurances, power consumptions and performance
• FPGA Offload （Tencent）• Pictures and videos compression and decompression adopt FPGA
• Scorpio Project• Compute resource and storage resource dis-aggregation
55
Win-Win Partnerships
LAB
OS/Virtualization
MW/DB
ISV
Industrial
customers
Core
components
Hangzhou, global
computing innovation center
56
Build standard-based open ecosystem
Application
OS
Virtualization
Server
Network
Storage
…
…
…
…
…
…
10,000+compatible products
1,000,000+compatible scenarios
500+vendor certifications
350+Huawei Ready
57
Joint innovation with open and collaboration
More than 500
business partners
100+ standard &
Technology organizations
THANK YOU
Copyright©2016 Huawei Technologies Co., Ltd. All Rights Reserved.
The information in this document may contain predictive statements including, without limitation, statements regarding the future financial and operating results, future product portfolio, new
technology, etc. There are a number of factors that could cause actual results and developments to differ materially from those expressed or implied in the predictive statements. Therefore, such
information is provided for reference purpose only and constitutes neither an offer nor an acceptance. Huawei may change the information at any time without notice.
Related Documents
