This presentation will take you through the underlying Intel technologies available in AWS EC2 Instances, what benefits these technologies can provide and how to use these technologies to maximise your use case workloads AWS EC2 Instances provide a wide selection of instance types, optimised to fit a broad and diverse set of use cases, all supported by the latest Intel Xeon processors and technologies providing you with confidence to choose an AWS EC2 instance type that best meet your performance needs for compute intensive, memory intensive, or IOPS intensive applications.
So, if you are interested in how to maximise your existing code optimisations and investments or thinking about migrating workloads to AWS come to this session.
Speaker: Peter Kerney, Lead Enterprise Architect, Cloud, SDI and NFV, Intel
3. ‹#›
Client Continuum
Internet of Things
Big Data
Cloud
Security
Energy
Retail
Smart Cities
Healthcare
Finance
Transport
>1 Zetabyte Internet Traffic3
>15 Bn Connected Devices2
>1 Bn More Netizen’s1
Infrastructure Enabling New Horizons
Tectonic
Shifts
New
Business +
Usage
Models
…by
2015
4. ‹#›
Infrastructure Enabling New Horizons
in 2015 3x more bits in digital universe than
stars in the physical universe
7.9 ZB
of Data In the world was
created in the last 2 years.
90%
Storage of a Smart City project
in China
200PB
People calling, texting, tweeting & browsing
on cell phones
>5 Billion
People calling, texting, tweeting & browsing
on cell phones
966 Exabytes
On-demand video traffic per month 1
million minutes of video on global IP
networks every second *
3B DVDs
The data deluge has arrived!
*Cisco Visual networking index, Feb ’13 http://arstechnica.com/business/2012/05/bandwidth-explosion-as-internet-use-soars-can-bottlenecks-be-averted
5. ‹#›
In Enterprise, some clear trends are shaping
Big Data
6%OF ENTERPRISES
MAKING DECISIONS
WITH BIG DATA
ANALYTICS1
Cloud
9%OF ENTERPRISE
WORKLOADS
RESIDE IN PUBLIC
CLOUD2
HPC
12%OF U.S.
MANUFACTURING
FIRMS USE HPC
CLUSTERS3
6. ‹#›
Big Opportunity: Extract Value from Data
= Big Challenge
Revenue
Growth
Cost
Savings
Margin
Gain
x
THINGS
50
Billion
DATA
35 ZB
VALUE
=
7. ‹#›
Big Opportunity: For Enterprises
Learn or Perish!
Source: Bain
Companies that use analytics best are…
Have top-quartile
financial results
Make decisions faster
than others
Execute on decisions
faster
Make data-driven
decisions
2X 5X 3X
more likely to
2X
8. ‹#›
Cloud Growth Drivers: Digital Services Economy
Virtuous Cycle Driving 20%+ Cloud CAGR through 20171
1 Source: Intel internal analysis
Cloud: 75% of Growth from
Consumer Services1
Consumer vs. Business Cloud
Hosting
SMB SaaS
2011 2012 2013 2014
Forecast
Search
Social Media
Shopping
Gaming
Business
Consumer
New Services Driving Explosion of Devices
New Devices Driving Demand for Servers
CPU Volume
~400 DEVICES
SMARTPHONES
1 SERVER1drives
~40 DEVICES
CONNECTED
FACTORY TOOLS
1 SERVER2drives
~100 DEVICES
MEDICAL WEARABLE
DEVICES
1 SERVER2drives
~20 DEVICES
DIGITAL SIGNS
1 SERVER2drives
9. ‹#›
Devices
Catalyzed by a Virtuous Business/Computing Cycle
A new approach is needed
Other brands and names are the property of their respective owners.
Datacenter Devices
Services
10. ‹#›
Which has Seismic Effects on the CIO
“Always on” availability
Extreme Flexibility & Service Agility
Public Cloud Economics
Compliant to regulatory requirements
CEO/CFO
EXPECTATIONS
Consumer-like expectations
Tech-savvy
“Outside” alternatives
Fastest services TTM
BUSINESS
EXPECTATIONS
11. ‹#›
IT: Period of Transformation
Computer-Centric Network-Centric Human-Centric
Focused on
Productivity
through automation Focused on
Cost Reduction
through connectivity
Focused on
Rapid Service Delivery
through cloud & devices
12. ‹#›
Ongoing Stress on Datacenter Operations
Network
2-3 weeks to provision
new services1
Storage
40% data growth CAGR,
90% unstructured3
Server
Average utilization <50%
despite virtualization4
New Challenges are coming….
13. ‹#›
Resource%Pool
Software Defined Infrastructure
Orchestration Software
Compute Storage Network
Automated, easy to
create new apps and
services
Intelligent provisioning
and resource allocation
Scalable, highly efficient
infrastructure
Power Performance Security Thermals Utilization Location Latency
Application A Application B Application C Application D
14. ‹#›
Re-architect the datacenter with cloud and software defined
infrastructure (SDI)
SDI
▪ Real-time enterprise with
full datacenter
orchestration
▪ Open standards and
API-based ecosystem
▪ Optimized resource
pooling, provisioning
management, and
service assurance
SDN/NFV
SDS / NVM
Non-volatile
memory tier
Composable
infrastructure
U
ERP
Private Cloud
Big Data
Hybrid Cloud
Technology Trends
Infrastructure
Deployment
DevOps
IoT
Visual / Voice
Analytics
IaaS
Traditional IT
PaaS
IaaS
ITSM
ITIL
PaaS
SaaS
Cloud
delivery
Infrastructure
telemetry
Silicon
photonics
15. ‹#›
A world where applications define the system
High degree of flexibility
Greater governance
Predictable /
proven needs
High rate
of change
Supply
Chain
Social
Enablement
Sales Automation
CRM
Mobile UI
Email
PayrollERP
Human
Resource
Finance
Systems
Data
Reporting
Application
Development
Environment
Big Data Insight
Regulatory
Compliance
Decision
Support
Enterprise
Communication
Classify your workloads to support diverse application needs
Performance for core business Go fast and adapt quickly
Go fast and be secureEfficiency for back office
Systems of
Record
Systems of
Differentiation
Systems of
Innovation
Systems of
Differentiation
16. ‹#›
Journey to cloud is then inevitable
A cloud strategy addresses real pressures IT organizations face
A Cloud Strategy
addresses these pressures
Balance CAPEX/OPEX to enable “invest for base,
partner for spikes” deployment model
Shorten time to value and secure deployments with
a software defined architecture (SDI)
Deliver services on-demand; partner with business
leaders on innovations that matter
Pressure to reduce cost
• Do more with less
• Deliver globally
• Deliver broader service types
Pressure to deliver faster
• Meet timelines at
business pace
• 24 x 7 x 365 availability
Pressure to secure and comply
• Enforce security and
maintain privacy
• Adhere to policies/
governance
17. ‹#›
Networking for the Cloud
2-3 Weeks1 to set up
new network
infrastructure…
SOFTWARE
DEFINED
NETWORKING
NETWORKING
TODAY
1 hour to set up networking
infrastructure!
1:%Source:%Intel%IT%internal%estimate
Agility
Opex/Capex
$
Days : To Install On-site Fixed
Function Appliance
Days : Get an Appliance On-Line
Hours : Load Applications
10 min: Cloud Server Purchased
20 min: Cloud Server VMs Instantiated
10 min: 2 Applications Loaded
20 min: 3 NFV applications purchased
(Virtual Firewall, Router, ADC)
18. ‹#›
Cloud Economics
Workload Performance Metrics Amazon’s TCO Analysis¹
Performance / TCO is the key metric
Hadoop Queries
Storage Capacity
Web Transactions / Sec
VMs per System
Servers
Networking
Power
Distribution &
Cooling
Power
1:%Source:%James%Hamilton,%Amazon%http://perspectives.mvdirona.com/2010/09/overallLdataLcenterLcosts/%Other%names%and%brands%may%be%claimed%as%the%property%of%others
19. ‹#›
Cloud Economics
Rack Scale Architecture Software Defined Infrastructure
Increase performance per TCO$ and capacity per IT$
1:%Intel%IT%estimates,%2014.%SDI%assumes%future%benefits.%Projections%are%extrapolations%from%that%actual%Intel%IT%data%on%estimated%future%cost%drivers.%Results%have%been%estimated%based%on%internal%Intel%analysis%and%are%provided%
for%informational%purposes%only.%Any%difference%in%system%hardware%or%software%design%or%configuration%may%affect%actual%performance%or%cost.%%%
Simplified Platform Management
Accelerated Service Delivery
Improved Capital & Operational
Efficiency
2012 2013 2014 2015 2016 2017
0%
20%
40%
60%
80%
100%
$/VMinstance
Intel IT projected¹ cost reduction ‘12-‘17
Support needed / VM HW requirement/ VM
21. ‹#›
Intel continuously investing in DC technologies
Source:%IDC%WW%Server%Tracker%&%Intel%
Enterprise
IT
8% CAGR
Cloud
SPs
>25% CAGR
Comms
SPs
>25% CAGR
High Performance
Computing
>20% CAGR
Servers
Network
Storage
Software
CPUs, Accelerators, Ethernet & Infiniband Switch Silicon
CPUs, Flash, Caching & RAID SW
Hadoop, Lustre File System, Datacenter Manager, Wind River OS
CPUs, Ethernet & Infiniband Controllers, Management SW
22. ‹#›
AWS keeping pace with Intel’s relentless
pursuit of Moore’s Law
Intel® Core™
Microarchitecture
Intel®
Microarchitecture
Codename
Nehalem
Intel®
Microarchitecture
Codename Sandy
Bridge
Intel®
Microarchitecture
Codename Haswell
Merom
New
Micro-
architecture
65nm
Penryn
New
Process
Technology
45nm
Nehalem
New
Micro-
architecture
45nm
Westmere
New
Process
Technology
32nm
Sandy
Bridge
New
Micro-
architecture
32nm
Ivy
Bridge
New
Process
Technology
22nm
Haswell
New
Micro-
architecture
22nm
Broadwell
New
Process
Technology
14nm
23
22
TOCK TICK TOCK TICK TOCK TICK TOCK TICK
The number of transistors incorporated in
a chip will approximately double every 24
months."
Gordon Moore, Former CEO & Intel co-founder
23. ‹#›
AWS keeping pace with Intel’s relentless
pursuit of Moore’s Law
AWS
EC2
C3
Intel® Core™
Microarchitecture
Intel®
Microarchitecture
Codename
Nehalem
Intel®
Microarchitecture
Codename Sandy
Bridge
Intel®
Microarchitecture
Codename Haswell
Merom
New
Micro-
architecture
65nm
Penryn
New
Process
Technology
45nm
Nehalem
New
Micro-
architecture
45nm
Westmere
New
Process
Technology
32nm
Sandy
Bridge
New
Micro-
architecture
32nm
Ivy
Bridge
New
Process
Technology
22nm
Haswell
New
Micro-
architecture
22nm
Broadwell
New
Process
Technology
14nm
23
22
TOCK TICK TOCK TICK TOCK TICK TOCK TICK
The number of transistors incorporated in
a chip will approximately double every 24
months."
Gordon Moore, Former CEO & Intel co-founder
24. ‹#›
AWS keeping pace with Intel’s relentless
pursuit of Moore’s Law
AWS
EC2
C3
AWS
EC2
C4
Intel® Core™
Microarchitecture
Intel®
Microarchitecture
Codename
Nehalem
Intel®
Microarchitecture
Codename Sandy
Bridge
Intel®
Microarchitecture
Codename Haswell
Merom
New
Micro-
architecture
65nm
Penryn
New
Process
Technology
45nm
Nehalem
New
Micro-
architecture
45nm
Westmere
New
Process
Technology
32nm
Sandy
Bridge
New
Micro-
architecture
32nm
Ivy
Bridge
New
Process
Technology
22nm
Haswell
New
Micro-
architecture
22nm
Broadwell
New
Process
Technology
14nm
23
22
TOCK TICK TOCK TICK TOCK TICK TOCK TICK
The number of transistors incorporated in
a chip will approximately double every 24
months."
Gordon Moore, Former CEO & Intel co-founder
25. ‹#›
Custom Intel Xeon processors for AWS
C4 = highest performing EC2 instances
PERFORMANCE
New EC2 Instances – C4
26. ‹#›
Full table of Intel technologies available in EC2
Amazon EC2 provides a wide selection of instance types optimized to fit different use cases.
AWS publish Intel processor models and enabled technologies that deliver
value and choice to end customers
The latest Intel Xeon processors are utilized, providing you with high performance and value, and the ability to choose Amazon EC2 instance types that best
meet your performance needs for compute intensive, memory intensive, or IOPS intensive applications.
AWS Instance
Type
Intel® process
Technology
Intel®process
technology
Intel® AVX
Intel® AES-NI
Intel® Turbo Boost
SSD Storage
Computer
optimized c4
Intel®Xeon
E5-2666 v3
22nm Haswell
AVX 2.0
EBS-OPT
by default
I/O
Optimized I2
Intel®Xeon®
E5-2670 v2
22nm
Ivy Bridge
I/O
Optimized I2
Intel®Xeon®
E5-2670 v2
22nm
Ivy Bridge
General Purpose
Balanced
M3
Intel®Xeon®
E5-2670 v2
22nm
Ivy Bridge
Burstable
T2
Intel®Xeon®
Family
GPU
G2
Intel® Xeon
E5-2670
32nm
Sandy Bridge
Storage
Optimized HS1
Intel®Xeon®
family
27. ‹#›
AWS end customer’s benefit
Novartis reduced cost by choosing premium instances
1 NIBR, of Novartis Pharmaceuticals ran Next Gen Sequencing, Imaging & Modeling & Sim techniques (specifically Virtual Screening w/CPU bound, low mem, low IO, network,
parallel Benchmarking SW ran same job many times (workload avg 32 secs on AWS cc2.8xlarge vs 1min 7 sec on AWS m1.large) revealed that best ROI was with cc2vspot
instances
Intel does not control or audit the design or implementation of third party benchmark data or Web sites referenced in this document. Intel encourages all of its customers to visit the
referenced Web sites or others where similar performance benchmark data are reported and confirm whether the referenced benchmark data are accurate and reflect performance
of systems available for purchase.
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 http://www.intel.com/performance
65%
Cost reduction1
Up to
0
18
35
53
70
Lower Performing Premium
Job runtime in seconds1
(lower is better)Sec
52%
Decrease in
run-time1
Up to
28. ‹#›
Intel & AWS big data case studies
More at http://aws.amazon.com/intel and http://aws.amazon.com/solutions/case-studies
Accelerating your adoption of Big Data solution
Learn what our customers are
able to do in 60 minutes
The opportunities exist for data to transform businesses
in fields as diverse as travel, genomics, e-commerce and
space exploration. They are created by the on-demand
availability of the high performance Intel Xeon E5
processor family in the Amazon Web Services cloud.
More at https://www.cloudinsights.com/Big-Data/
29. ‹#›
Key Takeaways
Transformation to cloud is critical
to speed business time to value,
deliver services on-demand,
and turn IT from business support
to business partner.
Cloud infrastructure enables IT to
seamlessly connect
multiple pools of resources, deliver
insights with higher scalability
and secure access to centralized data
Intel and AWS are
working jointly to help you
leverage the power
of the Cloud