View
799
Download
2
Category
Tags:
Preview:
DESCRIPTION
Kirk Skaugen Intel Datacenter and Connected Systems Group GM discusses the growth of data and connected devices through 2015. Kirk discusses how the industry is preparing for this growth and Intel's role.
Citation preview
Cloud 2015: The Road to 15 Billion Connected Devices
Kirk Skaugen Corporate Vice President, Intel Corporation
General Manager, Datacenter and Connected Systems Group
Intel's Vision This decade we will create and extend computing technology to
connect and enrich the lives of every person on earth.
More Devices
1. IDC “Server Workloads Forecast” 2009. 2.IDC “The Internet Reaches Late Adolescence” Dec 2009, extrapolation by Intel for 2015 2.ECG “Worldwide Device Estimates Year 2020 - Intel One Smart Network
Work” forecast 3. Source: http://www.cisco.com/assets/cdc_content_elements/networking_solutions/service_provider/visual_networking_ip_traffic_chart.html extrapolated to 2015
BY 2015…
More Users More Data
>1 Billion More Netizen’s
15 Billion Connected Devices
>1,000 Exabytes
Internet Traffic
PC Growth
0
100
200
300
400
500
'95 '97 '99 '01 '03 '05 '07 '09 '11* '13*
Millions Total Client
Source: IDC, *Forecast
PC Growth
0
100
200
300
400
500
'95 '97 '99 '01 '03 '05 '07 '09 '11* '13*
Millions Total Client
Source: IDC, *Forecast
1M PCs Shipped Everyday
The Ultrabook™ Beyond Thin and Light
CREATION TO EXPRESS PEACE OF MIND NOT NEEDING TO WAIT
ALWAYS AVAILABLE AT A PRICE THAT WORKS REFLECTION OF ME
0
20
40
60
80
100
2009 2010 2011 2012 2013 2014
670% GROWTH
Consumption & Creation
COMPUTING
Source: IDC Source: Cisco Visual Networking Index
Storage Capacity Shipped Exabytes
*Forecast
Forecast
1,300
0
500
1,000
1,500
2009 2010 2015
MB/Month
*
Average Traffic per
SMARTPHONE
2,311
0
1,000
2,000
3,000
2009 2010 2015
MB/Month
*
Average Traffic per
TABLET
6,522
0
5,000
10,000
2009 2010 2015
MB/Month
*
Average Traffic per
NOTEBOOK
New
INTELLIGENT CONNECTIONS Are Emerging …
New
INTELLIGENT CONNECTIONS Are Emerging …
New
INTELLIGENT CONNECTIONS Are Emerging …
HEALTHCARE MANUFACTURING
TRANSPORTATION
ENERGY
PC, CE, PHONES COMMUNICATIONS
RETAIL 2015: 15 Billion Connected Devices
Digital
Unintelligent
Secure
Sensors
Intelligent
Connected
Analytics
Managed
Device Evolution
Local
Predictive
Intelligent Systems
0
1
2
3
4
2011 2015
Billions
Source: IDC
>2X
Driving an explosion of
DATA
Source: http://www.cisco.com/assets/cdc_content_elements/networking_solutions/service_provider/visual_networking_ip_traffic_chart.html
150,000,000,000,000,000,000 Bytes*
Through 2009
245,000,000,000,000,000,000 Bytes*
In 2010 Alone
Source: http://www.cisco.com/assets/cdc_content_elements/networking_solutions/service_provider/visual_networking_ip_traffic_chart.html
Total Data Crossing the Internet
*Exabytes
IP Data GROWTH
2015 Internet Traffic
>1,000 EXABYTES
1995 2000 2005 2010 2015 Source: http://www.cisco.com/assets/cdc_content_elements/networking_solutions/service_provider/visual_networking_ip_traffic_chart.html extrapolated to 2015
The “Big Data” Opportunity
Source: McKinsey Global Institute Analysis, Pikes Research
30B
40B Photos Hosted
Corporate Data Growth
US Industries
235TB/Corporation U.S. Library of Congress
2011
12M 2021
209B
Air Travel
240 TBs Generated per Cross-Country Flight
2011
130M 2015
~340M
Smart Meters RFID Tag Sales
48 Hours of Video Uploaded Every Minute
Pieces of Content per Month
We Have Made Tremendous Progress…
In The Last ~5 Years Alone:
20x Performance/Watt
Half The Platform Idle Power
15:1 Consolidation
Is the Infrastructure Ready?
5000x faster & 100,000x cheaper from Intel’s 4004 introduced in 1971 compared to the 32nm Westmere where chip speed has gone up by 5000x and transistors have become 100,000x cheaper adjusted for inflation (Intel TMG Sept 2010). See back up foils for performance and power comparison .
Performance tests and ratings are measured using specific computer systems and/or components and reflect the approximate performance of Intel products as measured by those tests. Any difference in system hardware or software design or configuration may affect actual performance. Buyers should consult other sources of information to evaluate the performance of systems or components they are considering purchasing. For more information on performance tests and on the performance of Intel products, visit Intel Performance Benchmark Lmitations
Infrastructure Must Evolve to Address IT Challenges
But Not Enough …
1. IDC Market Analysis, January 2010. 2. Source information in backup
* Other names and brands may be claimed as the property of others.
Today’s Technology Would Require
Building 45 New Coal Power Plants to
Support 2015 IT Infrastructure2
Efficiency
70% of Respondents Saying Security is Top
Concern In Moving to Public Cloud1
Security
IT will spend ~$2T on deployment and
operations thru 2015 unless smarter
infrastructure radically simplifies
management of virtualized environments.
Manageability
“We have seen lock-in return as a top concern …
routinely seeking alternatives to proprietary
virtualization and cloud computing technology “
August 2010
Lock-In
Transforming Industry Economics
~1 TFLOP
~$55K/GFLOP
<$100/GFlop
>500 TFLOPS
Performance $/GFLOP
Annual Server Unit Shipments Supercomputing in 1997 Supercomputing in 2010
1995 2000 2000 2005 2010 2015 1995 2000 2005 2010 1995
The Compute Model Evolution
Researchers
Many
Ubiquitous
Pervasive
Mainframes Client/Server Web Cloud
Cloud Computing Can Once Again Transform the Economics of Computing and Services
Save 45 Giga Watts in 2015
Accelerate $100B of Industry Services Through 2015
Save $25B of IT Spend in 2015
Sources: Save 45GW: http://www.epa.gov/RDEE/energy-resources/refs.html#coalplant and internal analysis
Save $25B of IT spend: Bain, Acelerate $100B of industry services through 2015: IDC and internal analysis
ONE YEAR AGO …
AUTOMATED IT can focus more on
innovation and less on
management
FEDERATED Share data securely
across public and
private clouds
CLIENT AWARE Optimizing services based
on device capability
Desktops Laptops Intelligent Systems Smartphones Tablets Smart TVs Netbooks
Intel’s Cloud 2015 Vision
Catalyst for Change
Open & Interoperable Solutions Essential
* Other names and brands may be claimed as the property of others.
Industry Standard Solutions
300+ IT Leaders Representing
$100B+ in Annual IT Investment
June, 2011: 1st IT Cloud Requirements
September, 2011: First POC Solutions
>300 GLOBAL IT LEADERS
AIMS Data Centre
SDN BHD
Getronics NL BV
Biznet Networks
Connectria Hosting
JARING Communications
Sdn Bhd
RampRate
Scope Infotech, Inc.
Temperature Control
STEERING COMMITTEE
CONTRIBUTING MEMBERS
ADOPTER MEMBERS
Intel Serves as Technical Advisor to the Alliance
Intel® Cloud Builders
* Other names and brands may be claimed as the property of others.
Solutions to Make it Easier to Build and Optimize Cloud Infrastructure
Infrastructure as a Service / Cloud Resource Mgmt
Cloud Storage/ Networking
www.intel.com/cloudbuilders
Cloud Security
Cloud Efficiency
Client-Aware
D1D Oregon
D1C Oregon
Fab 32 Arizona
Fab 12 Arizona
Fab 28 Israel
New Capacity for 14nm and Beyond
22nm Fab Upgrades
D1X Oregon
Development Fab
Fab 42 Arizona
High Volume Fab
Future Cost-Effective Services Rely on
MOORE’S LAW …
Energy-Efficient Performance Built on Moore’s Law
Source: Intel
32nm 45nm 1x
0.1x
0.01x
0.001x
65nm 22nm
Low
er
Tra
nsi
stor
Leakage
Higher Transistor Performance (Switching Speed)
* Projected
14nm*
22 nm Benefits Smallest Handhelds to Powerful Cloud-based Servers
Predictable Intel® Xeon® Execution: The Tick-Tock Model
Tick Tock Tick Tock
45nm 32nm
Westmere
22nm
Tick Tock
Ivybridge Haswell
Delivering Sustained Microprocessor Leadership
Tick Tock
65nm
Sandybridge Harpertown Woodcrest Nehalem Clovertown
Engineered Systems
Oracle Exalogic Elastic Cloud Oracle Database
Appliance
Oracle Exalytics Business Intelligence
Machine
Oracle Exadata Database Machine
Optimizing Java™ – Intel®’s Long Term Contributions
>15 Years of Experience Helping Java Run Best On Intel® Architecture
Intel® Architecture + Oracle SW = World Record Java Performance
Record Performance
SPECjbb2005 2,798,763 BOPS
Cisco UCS – B440 M2
4-socket
SPECjEnterprise2010 26,118.67 EjOPS
Oracle WebLogic Server Rel 10.3.5
Oracle Java HotSpot Server VM 1.6.0_26
(2) Cisco UCS – B440 M2
SPECjbb2005 1,408,935 BOPS
Cisco UCS – B230 M2
2-socket
Powerful. Intelligent
.
Powerful. Intelligent
.
Powerful. Intelligent
.
All testing was performed by Cisco and results are as of 10/4/2011. :
- SPEC and the benchmark names SPECjbb2005 and SPECjEnterprise2010 are trademarks of the Standard Performance Evaluation Corporation. For more information, see http//www.spec.org and http://www.cisco.com/go/ucsatwork.
- Configurations: The Cisco UCS B440 M2 that achieved 2,798,763 SPECjbb2005 bops at 699,691 bops/JVM is based on Intel® Xeon® processor E7-4800 product family. The Cisco UCS B230 M2 that achieved 1,408,935 SPECjbb2005 bops at 704,468 bops/JVM is based on
Intel® Xeon® processor E7-2800 product family.
- Software and workloads used in performance tests may have been optimized for performance only on Intel microprocessors. Performance tests, such as Specjbb2005 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.
- Other names and brands may be claimed as the property of others.
Memory Caching in the Cloud with Memcached A high-performance, open source, distributed memory object caching system
Web Servers Apache, Nginx,
Lighttpd
Apps Servers PHP, Java, Rails, C, Perl, Python
Cached Servers
MEMCACHED
Database MySql, PostgreSOL
PROXY
LOAD BALANCER
CLOUD
Broadest Mission
Critical Vendor Base
Advanced
Data Protection
and Reliability
World-Class
Economics
The Intel® Xeon® Processor E7 Family Accelerating Mission Critical Transformation
28 New World Records and Up to 40% Increase vs. Xeon® 7500
1
1 SPECint*_rate_base2006 benchmark comparing next generation Intel® Xeon® processor E7-4870 (30M cache, 2.40GHz, 6.40GT/s Intel® QPI) scoring 1,010 (includes Intel Compiler XE2011 improvements accounting for
about 11% of the performance boost) to X7560 (24M cache, 2.26GHz, 6.40GT/s Intel QPI) scoring 723 (Intel Compiler 11.1). Source: Intel SSG TR#1131.
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.
Oracle Exalytics Business Intelligence Machine
Optimized Hardware
Processors: 40 Intel® Xeon® E7 Processor Cores
Complete Data Center
PRODUCT REFRESH
Xeon 5600 (Westmere- EP)
Xeon E7 (Westmere-EX)
Xeon E3 (Bromolow)
Xeon E5 (Sandy Bridge-EP)
Spanning a Diverse Set of Workloads
Knights Ferry (MIC Software Platform)
The Future Intel® Xeon® Processor E5 Codenamed Sandy Bridge-EP
Powerful. Intelligent.
Efficient I/O • Integrated PCIe reduces latency and power
• Platform includes integrated 6Gb SAS for high performance local memory
Growing Performance • Dual processing up to 8 cores per socket hyper-threaded
• Up to 2X FLOPS with Intel® Advanced Vector Extensions
Advanced Security • Support for the latest Intel security features like Intel® Trusted Execution
Technology and Intel® AES New instructions
The Foundation of the Next Generation Datacenter
Xeon 5500 Xeon E5
Cloud and Supercomputing Drive Unprecedented
Early Ramp
Intel® Xeon® E5: Broadest Xeon Product Line
Expect to Launch Almost 2X the Designs of
Xeon 5500/5600
Xeon 5500 Xeon E5
2X 20X
Shipping Now to HPC & Cloud End-Users. Full Launch Early 2012.
Summary
• Cloud 2015: 2.5B people and 15B devices
• Embedded devices transforming to intelligent and connected
• Drives 1000 Exabytes/year over the internet
• Client & cloud future relies on Moore’s law fundamentals
• Oracle and Intel delivering world record performance
World Record Backup World Records (Overall, x86, or by socket count) on Intel® Xeon® processor E7 family (sorted by benchmark):
1. 4-socket (4S) World Record TPC Benchmark* C (TPC-C) claim based on published result using IBM System x* 3850 X5 server compared with all other 4S results. For more information, see http://www.tpc.org/tpcc/reseults/tpcc_perf_results.asp.
2. 2-socket (2S) Performance and Price/Performance World Record TPC Benchmark* E (TPC-E) claim based on published result using IBM System x*3690 X5 server compared with all other 2S performance results and all published price/performance results. For more information, see http://www.tpc.org/tpce/results/tpce_perf_results.asp.
3. 4-socket (4S) World Record TPC Benchmark* E (TPC-E) claim based on published result using IBM System x*3850 X5 8P server compared with all other 4S results. For more information, see http://www.tpc.org/tpce/results/tpce_perf_results.asp.
4. 8-socket (8S) Overall World Record TPC Benchmark* E (TPC-E) claim based on published result using IBM System x*3850 X5 8P server compared with all other results. For more information, see http://www.tpc.org/tpce/results/tpce_perf_results.asp.
5. 4-socket (4S), non-clustered World Record columnar database TPC Benchmark* H@300GB scale factor claim based on published result using Dell PowerEdge* R910 server compared with all other non-clustered results. For more information, see http://www.tpc.org/tpch/results/tpch_perf_results.asp?resulttype=noncluster&version=2%&currencyID=0.
6. 4-socket (4S), non-clustered World Record columnar database TPC Benchmark* H@1000GB scale factor claim based on published result using Dell PowerEdge* R910 server compared with all other non-clustered results. For more information, see http://www.tpc.org/tpch/results/tpch_perf_results.asp?resulttype=noncluster&version=2%&currencyID=0.
7. 8-socket (8S) Non-Clustered, Microsoft SQL Server* World Record TPC Benchmark* H (TPC-H) @ 1000GB scale factor based on published result using HP ProLiant* DL980 G7 server compared with all other 1000GB scale factor, non-clustered SQL Server* results. For more information, see http://www.tpc.org/tpch/results/tpch_perf_results.asp?resulttype=noncluster.
8. 2-socket (2S) x86 Record SAP SD* 2-tier claim based on published result using IBM System x*3690 X5 server compared with all other 2S x86 results. For more information, see http://www.sap.com/solutions/benchmark/sd.epx. Certification #2011032.
9. 4-socket (4S) x86 Record SAP SD* 2-tier claim based on published result using IBM System x* 3850 X5 server compared with all other 4S x86 results. For more information, see http://www.sap.com/solutions/benchmark/sd.epx. Certification #2011015.
10. 8-socket (8S) x86 Record SAP SD* 2-tier claim based on published result using IBM System x* 3850 X5 8P server compared with all other 8S x86 results. For more information, see http://www.sap.com/solutions/benchmark/sd.epx. Certification #2011034.
11. NEW! 2-socket (2S) Overall World Record SAP* Transaction Banking benchmark claim based on published result using IBM System x* 3690 X5 compared with all other results. For more information, see http://www.sap.com/solutions/benchmark/trbk.epx. Certification #2011035.
12. NEW! 8-socket (8S) Overall World Record SAP SD-Parallel* claim based on published result using Oracle Sun Fire* X4800 M2 server compared with all other results. For more information, see http://www.sap.com/solutions/benchmark/sd.epx. Certification #2011040.
13. 2-socket (2S) x86 Record SPECfp*_rate_base2006 Score claim based on published result using Cisco UCS* C260 M2 server compared with all other 2S x86 results. For more information, see http://www.spec.org/cpu2006/results.
14. 4-socket (4S) x86 Record SPECfp*_rate_base2006 Score claim based on published result using Dell PowerEdge* R910 server compared with all other 4S x86 results. For more information, see http://www.spec.org/cpu2006/results.
15. NEW! 8-socket (8S) x86 Record SPECfp*_rate_base2006 Score claim based on published result using Hewlett-Packard ProLiant* DL980 G7 server compared with all other x86, 8-socket results. For more information, see http://www.spec.org/cpu2006/results.
16. 128-socket (128S) Overall World Record SPECfp*_rate_base2006 Score claim based on published result using SGI Altix* UV1000 server compared with all other results. For more information, see http://www.spec.org/cpu2006/results.
17. 2-socket (2S) x86 Record SPECint*_rate_base2006 Score claim based on published result using Cisco UCS* C260 M2 server compared with all other 2S x86 results. For more information, see http://www.spec.org/cpu2006/results.
18. 4-socket (4S) x86 Record SPECint*_rate_base2006 Score claim based on published result using Cisco UCS* C460 M2 server compared with all other 4S x86 results. For more information, see http://www.spec.org/cpu2006/results.
19. 8-socket (8S) x86 Record SPECint*_rate_base2006 Score claim based on published result using HP ProLiant* DL 980 G7 server compared with all other 8S x86 results. For more information, see http://www.spec.org/cpu2006/results.
20. 128-socket (128S) World Record SPECint*_rate_base2006 Score claim based on published result using SGI Altix* UV1000 server compared with all other results. For more information, see http://www.spec.org/cpu2006/results.
21. 2-socket (2S) World Record SPECjbb*2005 claim based on published result using Cisco UCS* B230 M2 server compared with all other 2S results. For more information, see http://www.spec.org/jbb2005/results/.
22. 4-socket (4S) World Record SPECjbb*2005 claim based on published result using NEC Express*5800/A1080a-S server compared with all other 4S results. For more information, see http://www.spec.org/jbb2005/results/.
23. 8-socket (8S) World Record SPECjbb*2005 claim based on published result using HP ProLiant* DL980 G7 server compared with all other 8S results. For more information, see http://www.spec.org/jbb2005/results/.
24. 64-socket (64S) World Record SPECjbb*2005 claim based on published result using SGI Altix* UV1000 server compared to all other results. For more information, see http://www.spec.org/jbb2005/results/.
25. 4-socket (4S) World Record single-node SPECjEnterprise*2010 claim based on published result using Dell PowerEdge* R910 server compared with all other single-node results. For more information, see http://www.spec.org/jEnterprise2010/results/jEnterprise2010.html.
26. NEW! 4-socket (4S) World Record two-node SPECjEnterprise*2010 claim based on published result using Cisco UCS B440 M2 Blade server compared with all other two-node results. For more information, see http://www.spec.org/jEnterprise2010/results/jEnterprise2010.html.
27. 2-chips World Record SPECompL*base2001 claim based on published result using Cisco UCS B230 M2 reference server with all other 2-chips base results. For more information, see http://www.spec.org/omp2001/results/ompl2001.html.
28. 2-chips World Record SPECompM*base2001 claim based on published result using Cisco UCS B230 M2 reference server with all other 2-chips base results. For more information, see http://www.spec.org/omp2001/results/ompm2001.html.
29. 4-chips World Record SPECompL*base2001 claim based on published result using Cisco UCS C460 M2 reference server with all other 4-chips base results. For more information, see http://www.spec.org/omp2001/results/ompl2001.html.
30. 4-chips World Record SPECompM*base2001 claim based on published result using Cisco UCS C460 M2 reference server with all other 4-chips base results. For more information, see http://www.spec.org/omp2001/results/ompm2001.html.
31. 2-socket (2S) World Record SPECvirt_sc*2010 claim based on published result using IBM Bladecenter* HX5 with MAX5 server compared with all other 2S results. For more information, see http://www.spec.org/virt_sc2010/results/.
32. 4-socket (4S) World Record SPECvirt_sc*2010 claim based on published result using Hewlett-Packard ProLiant* DL580 G7 server compared with all other 4S results. For more information, see http://www.spec.org/virt_sc2010/results/.
33. 8-socket (8S) Overall World Record SPECvirt_sc*2010 claim based on published result using IBM System x* 3850 X5 8P server compared with all other 8S results. For more information, see http://www.spec.org/virt_sc2010/results/.
34. 256-socket (256S) World Record STREAM claim based on published result using SGI Altix* UV1000 server compared to all other results. For more information, see http://www.cs.virginia.edu/stream/top20/Bandwidth.html.
35. 2-socket (2S) World Record VMmark* 2 claim based on published result using Hewlett-Packard ProLiant* BL620c G7 server compared with all other 2S matched-pair results. For more information, see http://www.vmware.com/a/vmmark/.
36. 4-socket (4S) World Record VMmark* 2 claim based on published result using Fujitsu PRIMERGY* RX600 S6 server compared with all other 4S matched-pair results. For more information, see http://www.vmware.com/a/vmmark/.
37. NEW! Overall World Record VMmark* 2 claim based on published result using Cisco UCS* C460 M2 server in a 4-host, uniform configuration compared with all other results. For more information, see http://www.vmware.com/a/vmmark/.
Recommended