AWS re:Invent Comes to London 2019 - Keynote

© 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.
COMESTOLONDON
Quick business update

COMESTOLONDON
2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018
Millions of active customers

COMESTOLONDON
Largest number of startup customers

COMESTOLONDON
Largest number of enterprise customers

COMESTOLONDON
Largest number of public sector customers

COMESTOLONDON
Broadest ecosystem of system integrators: premier consulting partners

COMESTOLONDON
Broadest ecosystem of ISVs and SaaS providers

COMESTOLONDON
4 6 %
Y/Y growth
Q3 2018 vs. Q3 2017
$2 7 B
Revenue run rate
ANNUALIZ E D FROM Q3 2018

COMESTOLONDON
State of the cloud
Market Share Analysis: IaaS and IUS, Worldwide, 2017, 28 June
2018, Analyst(s): Colleen Graham, Sid Nag,
Ed Anderson, David Edward Ackerman, Fred Ng
AWS /
51.80%
Microsoft / 13.30%
Alibaba / 4.60%
Google / 3.30%
IBM / 1.90%
Other vendors /
25.0%
W o r l d w i d e m a r k e t s e g m e n t s h a r e

COMESTOLONDON
What do builders want?

COMESTOLONDON

COMESTOLONDON
Security
CERTIFICATIONS KEY MANAGEMENTENCRYPTION

COMESTOLONDON
116 ENCRYPTION
ENABLED SERVICES
52 SERVICES
INTEGRATED WITH KEY
MANAGEMENT SERVICE
Encryption &
key
management
Certifications
100% AWS SERVICES
ATTESTED AND GDPR
READY
PCI-DSS, HIPAA/HITECH,
FedRAMP,
FIPS 140-2, and NIST 800-171
COMPLIANCE
CERTIFICATIONS
RESOURCE & USAGE AUDITING
Threat detection and application security
FINE-GRAINED ACCESS
CONTROL
AWS Identity and Access
Management
ASSESSMENT & REPORTING
AWS Inspector
CONFIGURATION
COMPLIANCE
AWS Artifact
THREAT DETECTION
Amazon GuardDuty
DDOS PROTECTION
AWS Shield
MACHINE LEARNING-POWERED
SECURITY
Amazon Macie
WEB APPLICATION
FIREWALL
AWS Web Application Firewall
- WAF
AWS
CloudTra
il
AWS Trusted
Advisor
Amazon
CloudWatch
Security

COMESTOLONDON
RELATIONAL
DATABASE
MIGRATION
SERVICES
NON-RELATIONAL
DATABASE
Databases

COMESTOLONDON
Relational databases Non-relational databases
Amazon
RDS
Amazon
Aurora
Amazon DynamoDB
Amazon Neptune
Amazon ElastiCache
for Redis
Amazon ElastiCache
for Memcached
AWS Migration
Hub
Migration services
AWS Database
Migration
Service
Databases
O r a c l e
M i c r o s o f t
S Q L S e r v e r

COMESTOLONDON
Compute
INSTANCES SERVERLESSCONTAINERS

COMESTOLONDON
DEEPLY
INTEGRATED
CONTAINER
ORCHESTRATION
SERVERLESS
CONTAINERS
Containers
MANAGED
KUBERNETE
S 15 minute execution time
Stream processing
OSS Application framework
46 supported event connections
Lambda
EVENT-DRIVEN
SERVERLESS COMPUTING
API end points
Workflow orchestration
Distributed training
CONTAINER
IMAGE
REPOSITORY
BATCH
PROCESSING
175 INSTANCE TYPES
Hibernate
Largest in-memory instances (Up to
12TB, SAP-certified)
Instances
Most powerful GPU instances for
machine learning (P3dn)
100 Gbps performance for
HPC, machine learning,
and big data
Compute
FPGA instance (F1)
AMD (M5, R5, T3 families)
New A1 arm instances with
AWS processors
EC2 CAPABILITIES
EC2 Fleet
Lightsail
Elastic GPUs
Spot

COMESTOLONDON
Storage
BLOCK
STORAGE
O B J E C T
S T O R A G E
F I L E
S T O R A G E
D ATA
T R A N S F E R

COMESTOLONDON
AWS Direct
Connect
AWS
Snowball
AWS Snowball
Edge
Provisioned
IOPS SSD
General
Purpose SSD
Throughput-
Optimized HDD Cold HDD AWS
Snowmobile
AWS Storage
Gateway
Amazon
Kinesis
Firehose
F i l e
D a t a
Tr a n s f e r
B l o c k
THE MOST VOLUME OPTIONS (4)
Amazon
Kinesis Data
Streams
Amazon
Kinesis Video
Streams
Amazon S3
Transfer
Acceleration
AWS
DataSync
AWS
Transfer for
SFTP
N E W N E W
Amazon Data
Lifecycle Manager
Elastic Volumes
O b j e c t

COMESTOLONDON
Amazon
S3

COMESTOLONDON
Amazon
S3
Best security, compliance,
and audit capabilities
Auditing with CloudTrail Data Events to track how, when,
and who is using individual objects in storage
S3 Inventory report for daily inventory report with encryption
status for objects in a bucket
Access control using customer-defined tags on individual objects
Write-once-read-many (WORM) controls on individual objects
S3 Block Public Access prevents public access for all current
and future storage
Use ML-powered Amazon Macie to automatically discover,
classify, and protect sensitive data
Unmatched durability,
availability, and scalability
All S3 and Glacier 3-AZ storage classes
replicate to 3+ AZs in a single region
Cross-region replication policies support any
AWS region and storage class as destination
Fine-grained control of cross-region
replication at bucket, prefix group, or object
tag level
Storage Class Analysis for lifecycle policy
setting recommendation
Amazon S3 Intelligent-Tiering
Amazon S3 Standard
Amazon S3 Standard-IA
Amazon S3 One Zone-IA
Amazon S3 Glacier (Unification)N E W
N E W
Easiest to use storage classes
Extract data from an object
with
S3 Select
Replicate objects by tag
with Cross-Region
Replication (CRR)
Audit access to objects
with CloudTrail Data
Events
Tier using lifecycle by tag
Set object-level access
control and security policies
using tags
Apply retention policies to
objects (WORM)
View operational metrics by
tag (CloudWatch)
Batch operations on millions
or billions of objects
N E W
Most object-level controls

COMESTOLONDON
Introducing Glacier Deep Archive
No tape to manage
C O M I N G I N 2 0 1 9NEW!
L o w e s t c o s t s t o r a g e a v a i l a b l e i n t h e c l o u d … e v e n l o w e r t h a n o n - p r e m i s e s t a p e
Designed for
99.999999999% durability
Recover data in hours vs.
days/weeks
$0.00099/GB/month
Less than 1/4th the cost of
Glacier

COMESTOLONDON
E a s i e s t w a y t o r u n s t a n d a r d f i l e s y s t e m s i n t h e
c l o u d
Amazon Elastic File System
Elastically scales up
and down with no
provisioning
Redundantly
stored across 3
Availability
Zones
4 performance modes
with General Purpose,
Max I/O, Burst, and
Provisioned Throughput
New EFS IA storage class
saves up to 85% on
infrequently accessed files
C O M I N G S O O N

COMESTOLONDON
Tens of thousands of customers using EFS

COMESTOLONDON
What about Windows
and file systems?

COMESTOLONDON
AWS /
57.70%
Microsoft Azure
/ 30.90%
Others /
11.40%
Worldwide Windows
Public Cloud IaaS
Instances by
Cloud Provider

COMESTOLONDON
Windows native for fully compatible Windows
File System experience
No hardware or software to
manage
Secure and compliant including
PCI-DSS, ISO, and HIPAA
Up to 10s of GB/s throughput with
sub-millisecond latencies
(Compatibility with AD, Windows access control, and
native Windows Explorer experience)
Amazon FSx for
Windows File
Server
F u l l y m a n a g e d W i n d o w s f i l e
s y s t e m b u i l t o n n a t i v e W i n d o w s
f i l e s e r v e r s

COMESTOLONDON
Is that ALL that’s needed in file systems?

COMESTOLONDON
High throughput, low latency –
100s of GB/s and millions of IOPS
Seamless integration with Amazon S3
Secure and compliant including PCI-
DSS, ISO, and HIPAA
F u l l y m a n a g e d f i l e s y s t e m f o r
c o m p u t e - i n t e n s i v e w o r k l o a d s
Amazon FSx
for Lustre

COMESTOLONDON
AWS Direct
Connect
AWS
Snowball
AWS Snowball
Edge
Provisioned
IOPS SSD
General
Purpose SSD
Throughput-
Optimized HDD
Cold
HDD AWS
Snowmobile AWS Storage
Gateway
Amazon
Kinesis
Firehose
F i l e
D a t a
Tr a n s f e r
B l o c k
THE MOST VOLUME OPTIONS (4)
Amazon EFS
Amazon FSx for
Windows File
Server
THE MOST CLOUD NATIVE SYSTEMS (3)
N E W
Amazon FSx
for Lustre
N E W
Amazon EFS
Standard
Amazon
EFS IA
2 STORAGE CLASSES
N E W
Amazon
Kinesis Data
Streams
Amazon
Kinesis Video
Streams
Amazon S3
Transfer
Acceleration
AWS
DataSync
AWS
Transfer for
SFTP
N E W N E W
Amazon Data
Lifecycle Manager
Elastic Volumes
4 PERFORMANCE
MODES
Max I/O
Burst
Provisioned
Throughput
General Purpose
Redundantly stores
across at least 3
AZs
Elastically scales up
and down
Automated storage
tiering based on
access
Amazon S3
One Zone-IA
Amazon S3
Standard
Amazon S3
Intelligent-Tiering
O b j e c tSTORAGE CLASSES (6)
Amazon
Glacier
Amazon S3
Standard-IA
Amazon S3 Glacier
Deep Archive
N E W
N E W
Auditing with CloudTrail Data
Events to track how, when, and
who is using individual objects
in storage
S3 Inventory Report for daily
inventory report with encryption
status for objects in a bucket
Access control using customer-
defined tags on individual
objects
Write-once-read-many (WORM)
controls on individual objects
S3 Block Public Access
protects against unintended
release of data
Use ML-powered Amazon
Macie to automatically discover,
classify, and protect sensitive
data
BEST SECURITY,
COMPLIANCE, AND
AUDIT CAPABILITIES
UNIQUE ANALYTICS CAPABILITIES
S3 Select supports Parquet, bzip2, JSON Arrays (in
addition to CSV, JSON, GZIP)
S3 Select integration with EMR for Hive and Presto
S3 performance improvements to retrieve data for
querying
All S3 and Glacier 3-AZ storage classes replicate to 3+ AZs in a single
region
Cross-region replication policies support any AWS Region and storage
class as destination
Fine-grained control of cross-region replication at bucket, prefix group, or
object tag level.
UNMATCHED DURABILITY, AVAILABILITY, AND
SCALABILITY
MOST OBJECT-LEVEL
CONTROLS
Extract data from an object with S3
Select
Replicate objects with Cross Region
Replication (CRR)
Audit access to objects with CloudTrail
Data Events
Set object-level access control and
security policies using tags
Apply retention policies to objects
(WORM)
View operational metrics by tag
(CloudWatch)
Batch operations on millions or trillions
of objects
N E W

COMESTOLONDON
The right tools for every builder

COMESTOLONDON
Landing in the zone

COMESTOLONDON
Setting up a Landing Zone
How do I maintain security and
compliance as more of my teams
move to the cloud?
Are there any best practices
for setting up my multi-
account environment?
How can I set and
enforce policies for
all my workloads?
What AWS tools
should I use?

COMESTOLONDON
Dashboard for continuous visibility
Guardrails for policy enforcement
Automated Landing Zone
with best-practices blueprintsAWS Control Tower
T h e e a s i e s t w a y t o s e t u p a n d g o v e r n a
s e c u r e , c o m p l i a n t , m u l t i - a c c o u n t
e n v i r o n m e n t o r L a n d i n g Z o n e

COMESTOLONDON
Separating the signal
from the noise

COMESTOLONDON
View summary of prioritized
issues
Automate compliance checks to detect deviations
against industry standards
(e.g. Center for Internet Security AWS Foundations Benchmark)
Save time by aggregating
alerts
NEW!
C e n t r a l l y m a n a g e s e c u r i t y a n d
c o m p l i a n c e a c r o s s a n A W S
e n v i r o n m e n t
AWS Security Hub

COMESTOLONDON
AWS Security Hub partners

COMESTOLONDON
Jumping into the lake!

COMESTOLONDON
Set up
Storage
1
Move data2
Cleanse and
prep data
3
Configure and enforce security and
compliance policies
4
Make data accessible
for analytics
5
Steps for building and managing a data lake

COMESTOLONDON
Enforce security policies across
multiple services
Gain and manage new insights
Move, store, catalog, and clean your
data faster with machine learning
A service that allows you to build
a secure data lake in days
AWS Lake
Formation

COMESTOLONDON
London Borough of Waltham Forest:
Smart Data Lake
Richard Holland
Assistant Director Technology, Innovation Digital and ICT

COMESTOLONDON
Location

COMESTOLONDON
Summary
Waltham Forest is a predominantly residential borough
with one of the smallest economies in London (9% 2016 vs.
6% in London). Self-employment rate higher than London
(16.5% vs. 13.6%)
Strong business growth in recent years but less impact on
job growth as mainly small businesses
High population growth throughout last decade driven by
international migration (mainly from the EU) 275,800 in
2016
Young age structure with more children and working-age
residents compared to the UK average. Over 65,000 children
and young people (about a quarter of population)
Increasingly diverse population without a single majority
group
House prices have rocketed since the recession and the
borough is among the ten local areas in the country
where house price/earnings ratio has increased most
since 1999.
High population churn with more people moving to
other areas in the UK than other way around,
particularly families with young children. 42% of
outflows are to other London boroughs. 74% of inflows
are from other London boroughs, mainly neighbouring
boroughs.
Population is projected to continue to grow and is
increasingly ageing
Significant uncertainty around migration and the wider
impacts of Brexit
Ranked as 35th most deprived local authority in England
(out of 326) in 2015

COMESTOLONDON
Waltham Forest Services
Rubbish and Recycling
Planning & Building Control
Special Education Needs and Transport
Adult Social Care
Housing
Enforcement and Licensing
Arts, Part and Libraries
Benefits and Money Advice
Electoral Services
Births, Deaths & Marriages
Regeneration
Children Services (Fostering, Adoption,
Support)
Telecare
Commercial Trading Company (Pest Control,
Weddings etc.)
Highways & Parking
CCTV
Public Health
Our IT department consist of approximately 70 individuals supporting over 150 systems all in silos, more than 80
customer portals & websites, and one Chat Bot. We have loads of data! BUT It is very difficult to combine data
together and produce reliable , real time intelligence
We have 2800 officers and 60 Councillors running the following services for 280k residents.

COMESTOLONDON
Tenure Intelligence: What is the problem?
• Identify the tenure and ownership of private residential property.
• Identify potentially unlicensed properties and problem properties.
What do we want to know?
• Property related information is stored across multiple databases.
• Labour intensive back-off checks and analysis.
• Data is not always up-to-date accurate in regards to privately rented
properties.
What is the current approach?
• Better identification of unlicensed properties to support targeted
enforcement action.
• Better evidence base to track reductions in anti-social behaviour e.g. FPNs,
ASB.
• Robust single database with up-to-date information to monitor changes in
property tenure and problem property related issues.
Why use a Tenure Intelligence System?

COMESTOLONDON
How does Smart TIMS work?
• Profile and create sample Data Quality rules using Big Data Quality, show
how this insight can be exposed and leveraged by Waltham Forest
stakeholders via the Big Data, Data Quality software
• Combine the solution architecture, technical details, analytics, statistics etc.
into a final presentation
• Provide high-level walkthrough of the data journey, ingestion using S3, reuse
of components and APIs
• Overview and demonstration of Data Quality, creation of new business rule,
running the plan and examination of the result
• Using a Predictive Algorithm tells us where the private landlords are and who
is not performing.
TIMS
Data
Lake
Capita
Property
Data
Academy
Council Tax
Xpress
Electoral
Register
iWorld
Northgate
M3 Social
Housing
Civica
• Build mappings and ingest data to Hive with Big Data Manager (“BDM”)
• Provision data in Informatica Enterprise Data Catalog (“EDC”)

COMESTOLONDON
Predictive Tool
Statistical regression used to ‘weight’ different variables.
Regression model looks for links and changes between
variables that may indicate unlicensed HMOs.
Weighting transferred into algorithm to provide
unlicensed property risk-rating (0-100).
Measures relationship between variables and insight on
cause/effect.
The algorithm is embedded into the AWS data lake
environment and updated in real time.
Unlicensed
Residential
Property
Changes
in Council
Tax payer
Changes
to
Electoral
Role
Benefit
and
Tenure
Type
Anti Social
Behaviour
Pest
Control
Private
Housing
Notices
Balance of
Council
Tax
Payments

COMESTOLONDON
Risk based profile of landlords

COMESTOLONDON
Total per ward

COMESTOLONDON
SmartIntegrationHub
CouncilServices
Data Lake(s) Data
Preparation
and Advanced
Analytics
Dashboards
and
Visualisation
Smart Integrations – Potential
Potential
Rent Arrears Prediction
Real Time Social Economic Forecasts
Council Tax Arrears Prediction
Homelessness Indicators
Customer Profiles
Image Recognition and Analytics
Waste prediction and Fixed Penalty Notice Prediction
Social Listening and Real time Council Service Satisfaction
Planning and Areas of decline prediction
What ifs
Connect Voice and Contact Centre and start analysing true channel shift?
Smart Things, Sensors and GIS?
Public Health and other public sector bodies

Thank you!
COMESTOLONDON
Richard Holland
Assistant Director Technology, Innovation Digital and ICT
London Borough of Waltham Forest

COMESTOLONDON
A m a z o n
D y n a m o D B
A m a z o n
E l a s t i C a c h e
A m a z o n
N e p t u n e
K e y Va l u e D o c u m e n t I n - M e m o r y S t o r e G r a p h
Evolution of databases for all of your application needs
R e l a t i o n a l
A m a z o n
R D S
A U R O R A C O M M E R C I A
L
C O M M U N I T Y
O r a c l e
M i c r o s o f t
S Q L
S e r v e r
R e d i s
M e m c a c h e d

COMESTOLONDON
2017/10/09 21:45:15 6119836 5368
2017/10/09 21:45:15 6119836 5768
2017/10/09 21:45:15 6119836 4568
2017/10/09 21:45:15 6119836 5368
2017/10/09 21:45:15 6119836 4668
2017/10/09 21:45:16 6119836 2348
2017/10/09 21:45:16 6119836 0978
2017/10/09 21:45:16 6119836 2947
2017/10/09 21:45:16 6119836 5368
2017/10/09 21:45:16 6119836 2428
2017/10/09 21:45:16 6119836 8031
2017/10/09 21:45:17 6119836 1987
162471437243765 0.00349 0.125 22.0987
372431437243787 -0.0625 0.125 22.9870
243765562471437 0 0.125 22.7850
143724376516249 - 0.0625 0.125 22.6752
654714765372430 0.25 0.125 22.2974
247141637243765 0.1876 0.125 22.0918
162471437243765 0 0.125 22.1785
989272640098287 -0.0675 0.125 22.9836
143729624798927 0.25 0.125 22.7156
729621437243989 0.1876 0.125 22.0183
724376243765718 -0.0675 0.125 22.1930
437243765162471 0.25 0.125 22.8625
437243247124437 0 0.125 22.8154
Tracking change over time
85
87
89
91
93
95
5:28:15 PM 5:28:30 PM 5:28:45 PM 5:29:05 PM
Humidity
% WATER VAPOR
Timestamp angle torque speedLog
86.0
1
2
3
Clickstream data
IoT sensor readings
DevOps data

COMESTOLONDON
Existing time-series
databases
Relational databases
Difficult to scale
Manual effort needed for enterprise-
grade availability and reliability
Limited data lifecycle
management capabilities
Unnatural for time-series data
Rigid schema inflexible for fast-moving
time-series data
Building with
time-series
data is
challenging
Lack time-series analytic functions like
smoothing, approximation, and interpolation

COMESTOLONDON
NEW!
F a s t , s c a l a b l e , f u l l y
m a n a g e d t i m e - s e r i e s
d a t a b a s e
A V A I L A B L E I N P R E V I E W T O D A Y
Amazon
Timestream

COMESTOLONDON
Timestream is 1,000X faster and 1/10th
the cost of relational databases
Trillions of daily events
Serverless
Time-series analytics (interpolation, smoothing,
approximation) built in
Multiple orders of magnitude improvement in
query performance
F a s t , s c a l a b l e , f u l l y
m a n a g e d t i m e - s e r i e s
d a t a b a s e
Amazon
Timestream

COMESTOLONDON
How do we think about Blockchain?

COMESTOLONDON
TRANSACTIONS WITH
DECENTRALIZED TRUST2
DMV
Track vehicle title
history
Manufacturers
Track distribution of a recalled
product
HR & Payroll
Track changes to an
individual’s profile
Healthcare
Verify and track hospital
equipment inventory
Common customer use cases
Relational
databases
Blockchain
frameworks
Using sub-optimal ways
to solve the problem
LEDGERS WITH
CENTRALIZED TRUST
1

COMESTOLONDON
Mortgage
Lenders
Process syndicated loans
Small Businesses
Transact with suppliers and
distributers
Retail
Streamline customer
rewards
Financial
Institutions
Peer-to-peer payments
Blockchain Frameworks
Hard to scaleSet up is hard
Complicated
to manage
Expensive
TRANSACTIONS WITH
DECENTRALIZED
TRUST
2
LEDGERS WITH
CENTRALIZED TRUST
1
Common customer use cases

COMESTOLONDON
Two needs, what to do?

COMESTOLONDON
Cryptographically
Verifiable
All changes are cryptographically
chained and verifiable
Transparent
Full visibility into entire data lineage
Immutable
Append-only, immutable journal
tracks history of all changes
Highly scalable
Automatically scale up or down
Easy to use
Query with familiar SQL operators
Fast
Execute 2-3X more transactions
F u l l y m a n a g e d l e d g e r d a t a b a s e
w i t h a c e n t r a l t r u s t e d a u t h o r i t y
Amazon Quantum
Ledger Database
(QLD)

COMESTOLONDON
Choose Hyperledger Fabric or Ethereum
Create blockchain networks with a few clicks;
manage them with simple API calls
Scales to support thousands of
applications running millions of
transactions
Easy to move data into QLDB for further analysis
F u l l y m a n a g e d b l o c k c h a i n s e r v i c e ,
s u p p o r t i n g b o t h H y p e r l e d g e r F a b r i c a n d
E t h e r e u m f r a m e w o r k s
Amazon
Managed
Blockchain

COMESTOLONDON
Transforming UK Healthcare
with EMIS-X
Pete Malcolm
Group Chief Technology Officer
EMIS Group plc

COMESTOLONDON
• Major supplier of IT software and services to NHS and private
healthcare
• Multiple sectors:
- Primary care (GPSoC)
- Community care (Child and Mental Health)
- Acute care (Hospitals)
- Urgent care (A&E)
- Pharmacies (Community and Hospital)
- Diabetic Retinopathy
• Multiple Markets
- England
- Scotland
- Wales
- Northern Ireland
- Gibraltar
• Key Metrics:
• 40 million live patient records
• 2.2 billion patient documents
• 160 million appointments annually
• 640 million prescription annually
• 250 million transactions daily
• 157,000 desktop clients
• 2 data centres
• 680 physical servers
• 3,200 virtual machines
• 300 SQL Server instances
• 10,500 databases
• >1PB storage

COMESTOLONDON
• Multiple vendors with disparate systems
• Lack of federated working
• Interoperability limited to data exchange
• Multiple records per individual citizen
• Limited access to patient records causes clinical risk
• High barrier to new market entrants
• Inefficiency, duplication, manual processes
• Patient and clinical professional inconvenience
• Systemic risk due to lack of joined up services

COMESTOLONDON
• Cloud based clinical platform
• Hosted exclusively in UK data centres
• Extends managed services model with clinical modules
• Multi-tenant with controlled data sharing
• Foundation modules provide common services via API
• Application modules provide user apps via UX/UI
• Applications can be internal or external
• Supports third party applications and full interoperability

COMESTOLONDON
Person/patient record relationship
Patient record and permissions

COMESTOLONDON
EMIS solutions
• Provides rapid deployment of upgrades/ fixes
• Reduces development times
• Reduces operating costs
• Improves security, reliability and data resilience
• Demonstrates a progressive migration path from
EMIS Web
• Facilitates integration between EMIS Group
applications
• Provides practically infinite scalability and future-
proofing
Ecosystem
• Facilitates fully “joined-up” UK healthcare
• Multi-vendor and organisation participation
• Provides federated access to data
• Retains tenant data ownership
• Preserves third party commercial interests
• Significantly reduces barrier to market entry for third-party
technologies
• Improves patient outcomes, control and convenience
• Dramatically increases NHS efficiency and reduces cost

COMESTOLONDON
• Managed Services
• Significant reduction in development time and cost
• Improved time to market
• Reduced maintenance costs
• Serverless architecture eliminates OS maintenance and
downtime
• Physical Infrastructure
• Highest levels of security
• Built for purpose
• Massive redundancy
• Eliminates building issues
• Performance, Scalability, Reliability
• Event driven architecture provides real-time
response
• Full demand-based auto-scaling eliminates peak
demand performance reduction and minimises cost
• Cloud services scaling and capacity always
available – no expansion outages
• Full redundancy across three availability zones for
maximum resilience
• 99.999999999% data durability (equivalent to one
lost file every 659,000 years)

COMESTOLONDON
Current
• 8 EC2 instances
• S3 long term storage
• Separate search DB
• Manual configuration
• Separately managed
• Heavy maintenance load
• Complex to extend
AWS Managed Blockchain
• Dramatic savings in deployment,
configuration, management,
maintenance and expansion time
• Significantly reduced operational
risk
• Cost TBD

Thank you!
COMESTOLONDON
Pete Malcolm
Group Chief Technology Officer
EMIS Group plc

COMESTOLONDON
A lot of progress with
machine learning in the last 12 months

COMESTOLONDON
More machine learning is happening on AWS than anywhere else
10,000s
of active customers
2xcustomer references

COMESTOLONDON
More machine learning is happening on AWS than anywhere
else

COMESTOLONDON
How we think about
machine learning

COMESTOLONDON
F R A M E W O R K S
M L F r a m e w o r k s
+ I n f r a s t r u c t u r e
M L S e r v i c e s
A I S e r v i c e s
I N T E R F A C E S I N F R A S T R U C T U R E
P3 P3dn C5 C5n AWS
Greengrass
A m a z o n
S a g e M a k e r
Amazon
Transcribe
Amazon
Polly
Amazon
Lex
C H A T B O
T S
Amazon
Rekognition
Image
Amazon
Rekognition
Video
V I S I O N S P E E C H
Amazon
Comprehend
Amazon
Translate
L A N G U A G
E S

COMESTOLONDON
1 0 0 G b p s
n e t w o r k i n g
P3dn instances
3 2 G B G P U
m e m o r y
7 6 8 R A M ( S y s t e m
M e m o r y )
S c a l e m o d e l t r a i n i n g p e r f o r m a n c e a c r o s s m u l t i p l e i n s t a n c e s
P3dn.24xl
3 X f a s t e r n e t w o r k t h r o u g h p u t t h a n a n y
o t h e r p r o v i d e r
2 X a s m u c h G P U m e m o r y
t h a n a n y o t h e r G P U i n s t a n c e
f r o m o t h e r m a j o r p r o v i d e r s
1 0 0 + G B m o r e s y s t e m m e m o r y t h a n
t h e n e x t l a r g e s t i n s t a n c e a v a i l a b l e
f r o m o t h e r p r o v i d e r s
Infrastructure and frameworks for machine learning
Best price/performance
For model training
I N F R A S T R U C T U R E
P 3 P3dn
C O M I N G
S O O N

COMESTOLONDON
I N T E R F A
C E S
F R A M E W O R K S
Infrastructure and frameworks for machine learning
Best price/performance
For model training
I N F R A S T R U C T U R E
P 3 P3dn
C O M I N G
S O O N

COMESTOLONDON
In the cloud, 85% of TensorFlow workloads run on AWS
Nucleus Research, TensorFlow on AWS, 28 November
2018, Analyst: Rebecca Wettemann

COMESTOLONDON
Scaling TensorFlow
STOCK
TENSORFLOW
65%
scaling efficiency
with 256 GPUs

COMESTOLONDON
Scaling TensorFlow 65%
STOCK
TENSORFLOW
scaling efficiency
with 256 GPUs
90%
AWS-OPTIMIZED
TENSORFLOW
scaling efficiency
with 256 GPUs
N E W

COMESTOLONDON
Fastest
training
times for
TensorFlow
30m
training time
(Mountain View)
ResNet50 only
Specialized hardware
only available in beta
14m
training time
(Seattle)
ResNet50, convolutional neural
networks (for images), recurrent
neural networks (for language
recommendations)
Optimized for P3 with
global availability
N E W F A S T E S T
T R A I N I N G
T I M E

COMESTOLONDON
But what about
inference?
It’s never been easier, faster,
or more cost-effective to train
machine learning models

COMESTOLONDON
Running inference in production
drives the majority of cost
for machine learning Inference
Training
Infrastructure costs

COMESTOLONDON
Two main drivers of inference
inefficiency: complexity and cost of
machine learning inference today
Elasticity is
important
One size does
not fit all

COMESTOLONDON
NEW!
G E N E R A L L Y A V A I L A B L E T O D A Y
A d d G P U a c c e l e r a t i o n t o a n y A m a z o n E C 2
i n s t a n c e f o r f a s t e r i n f e r e n c e a t m u c h l o w e r
c o s t ( u p t o 7 5 % s a v i n g s )
Amazon
Elastic Inference
E C 2
In s ta n c e
E C 2
In s ta n c e
E C 2
In s ta n c e
G P U

COMESTOLONDON
Amazon
Elastic Inference:
how it works
P 3 . 8 X L
P 3 P 3
P 3 P 3
Amazon
Elastic Inference

COMESTOLONDON
3 6 T F L O P S
V P C
M 5 . l a r g e
Amazon
Elastic Inference
Amazon
Elastic Inference:
how it works

COMESTOLONDON
Amazon Elastic Inference: fast, low-cost machine learning inference
Starting at
1 TFLOPS
Any instance
family
Simple speech and
language models
Up to
32 TFLOPS
Recommendation engines
or fraud detection models
Provision Elastic
Inference capacity
inside VPC

COMESTOLONDON
360,000ResNet-50
Computer vision
deep learning model
images per hour,
inference
$0.22
per hour
on medium EI accelerator
75%
lower cost
Lowering inference costs with Amazon Elastic Inference
L O W E S T C O S T
A V A I L A B L E

COMESTOLONDON
Workloads needing entire GPU
or that are latency sensitive

COMESTOLONDON
NEW!
High performance machine learning inference chip,
custom designed by AWS
A V A I L A B L E L A T E 2 0 1 9
AWS Inferentia

COMESTOLONDON
AWS Inferentia: machine learning inference chip
High throughput
Low latency
Hundreds of TOPS
Multiple
data types
Multiple
ML frameworks
INT8, FP16,
mixed precision
TensorFlow, MXNet,
PyTorch, Caffe2, ONNX
EC2 instances
Amazon SageMaker
Amazon Elastic Inference

COMESTOLONDON
F R A M E W O R K S
M L F r a m e w o r k s
+ I n f r a s t r u c t u r e
M L S e r v i c e s
A I S e r v i c e s
I N T E R F A C E S I N F R A S T R U C T U R E
A m a z o n
S a g e M a k e r
Amazon
Transcribe
Amazon
Polly
Amazon
Lex
C H A T B O T S
Amazon
Rekognition
Image
Amazon
Rekognition
Video
V I S I O N S P E E C H
Amazon
Comprehend
Amazon
Translate
L A N G U A G E S
P3 P3dn C5 C5n Elastic inference Inferentia AWS Greengrass
N E WN E W

COMESTOLONDON
Amazon SageMaker
Bringing machine learning to all developers
Pre-built
notebooks for
common problems
C o l l e c t a n d
p r e p a r e
t r a i n i n g d a t a
B u i l t - i n , h i g h
p e r f o r m a n c e
a l g o r i t h m s
C h o o s e a n d
o p t i m i z e y o u r
M L a l g o r i t h m
One-click
training
Set up and manage
environments
for training
O p t i m i z a t i o n
F u l l y m a n a g e d
w i t h a u t o - s c a l i n g ,
h e a l t h c h e c k s ,
a u t o m a t i c
h a n d l i n g o f n o d e
f a i l u r e s , a n d
s e c u r i t y c h e c k s
S c a l e a n d
m a n a g e t h e
p r o d u c t i o n
Train and tune
model
(trial and error)
One-click
deployment
Deploy model
in production

COMESTOLONDON
More than ten thousand customers using Amazon SageMaker

COMESTOLONDON
Pre-built
notebooks for
common problems
C o l l e c t a n d
p r e p a r e
a l g o r i t h m s
C h o o s e a n d
One-click
training
Set up and manage
environments
for training
Train and tune
model
(trial and error)
Amazon SageMaker
One-click
deployment
h o s t i n g w i t h
a u t o - s c a l i n g
S c a l e a n d
m a n a g e t h e
p r o d u c t i o n
Deploy model
in production

COMESTOLONDON
Successful models require
high-quality training data

COMESTOLONDON
S T A T I O N A R Y C A R
S T A T I O N A R Y C A R M O V I N G C A R
P E D E S T R I A N
S T O P S I G N
D I R E C T I O N
S I G N
S T O P
S I G N
T R A F F I C
S I G N A L S
T R A F F I C
S I G N A L S
S T A T I O N A R Y V E G E T A T I O N
S T A T I O N A R Y R O A D
S T A T I O N A R Y B U I L D I N G
M O V I N G
C A R
P A V E M E N T
Successful models require
high-quality training data

COMESTOLONDON
NEW!
Build highly accurate training datasets and reduce data
labeling costs by up to 70% using machine learning
Amazon SageMaker
Ground Truth

COMESTOLONDON
Amazon SageMaker Ground Truth: how does it work?
Human
annotations
Data in S3
Automatic
annotations
Training
data
Simple, pre-built workflows
Mechanical Turk
Third-party vendors
Your own employees
Active
Learning
model
>80% confidence
<80% confidence

COMESTOLONDON
Train and tune
model
(trial and error)
Pre-built
notebooks for
common problems
a l g o r i t h m s
One-click
training
One-click
deployment
h o s t i n g w i t h
a u t o - s c a l i n g
S c a l e a n d
m a n a g e t h e
p r o d u c t i o n
C h o o s e a n d
Deploy model
in production
Set up and manage
environments
for training
Amazon SageMaker
C o l l e c t a n d
p r e p a r e
Amazon SageMaker
Ground Truth
N E W

COMESTOLONDON
Amazon SageMaker: broad set of built-in algorithms
Designed to
be 10x faster
K-Means Clustering
Principal Component Analysis
Neural Topic Modelling
Factorization Machines
Linear Learner (Regression)
BlazingText
Reinforcement learning
XGBoost
Topic Modeling (LDA)
Image Classification
Seq2Seq
Linear Learner (Classification)
DeepAR Forecasting
A L G O R I T H M S
Built into
Amazon SageMaker
Improving and
expanding continually
40% more algorithms added
since SageMaker launch

COMESTOLONDON
AWS Marketplace for
machine learning
Natural Language Processing
Computer Vision
Speech Recognition
Text Clustering
Text Generation
Text Classification
Grammar and Parsing
Named Entity Recognition
Text to Speech
Handwriting Recognition
Object Detection in Images
3D Images
Text OCR
Video Classification
Speaker Identification
Ranking
Regression
Anomaly Detection
Browse or search
AWS Marketplace
Subscribe
in a single click
Available through
Amazon SageMaker

COMESTOLONDON
Selling algorithms &
models on AWS
Marketplace
Register with AWS Marketplace
Automatically validate the algorithm or
model with a test run on SageMaker
Package algorithm, models,
and configuration
Self-service listing on AWS Marketplace

COMESTOLONDON
SOPHIST ICAT IO N
OF ML MODELS
Supervised
learning
A M O U N T O F T R A I N I N G D A T A R E Q U I R E D

COMESTOLONDON
SOPHIST ICAT IO N
OF ML MODELS
Unsupervised
learning

COMESTOLONDON
SOPHIST ICAT IO N
OF ML MODELS
Reinforcement
learning

COMESTOLONDON
How does reinforcement learning work?
A l g o r i t h m c o n t r o l s P A C - M A N
L e a r n t o p l a y t o g e t t h e h i g h e s t
s c o r e p o s s i b l e
M a x i m i z e r e w a r d s a n d m i n i m i z e
p e n a l t i e s
L e a r n a d v a n c e d s t r a t e g i e s

COMESTOLONDON
NEW!
New machine learning capabilities in Amazon SageMaker
to build, train, and deploy with reinforcement learning
Amazon SageMaker RL

COMESTOLONDON
R e i n f o r c e m e n t l e a r n i n g f o r
e v e r y d e v e l o p e r a n d d a t a
s c i e n t i s t
Fully managed reinforcement
learning algorithms
TensorFlow, MXNet,
Intel Coach, and Ray RL
2D and 3D simulation
environments via OpenGym
Simulate environments with
Amazon Sumerian and AWS
RoboMakerExample notebooks
and tutorials
Amazon
SageMaker RL

COMESTOLONDON
Call Centres in the Cloud
Amazon Connect lets you set up AI powered virtual call
centres with no infrastructure
Benedict Thompson Lead Developer

COMESTOLONDON
Difficult to administer and generate reports

COMESTOLONDON
Easy to use, administrator, and report

COMESTOLONDON
Horsham
Milton Keynes
Northampton

COMESTOLONDON
Amazon Lex AWS Lambda

COMESTOLONDON
Can we help developers get rolling
with reinforcement learning?
(literally)

COMESTOLONDON
NEW!
Fully autonomous 1/18th scale race car,
driven by reinforcement learning
A V A I L A B L E F O R P R E - O R D E R O N
A M A Z O N . C O M
AWS DeepRacer

COMESTOLONDON
HD Video Camera
Gyroscope
for direction and orientation
Accelerometer
for measuring change in speed
Two batteries:
one to power on-board
compute, one to drive motors
Dual-core
Intel Atom® Processor
Introducing AWS DeepRacer
All wheel drive,
monster truck chassis
Suspension
mounted high for a view of the road
Both accelerometer and gyroscope useful in the
future for building more sophisticated models such
as finding the perfect racing line or path finding

COMESTOLONDON
AWS DeepRacer: how does it work?
3D simulator with
virtual car and track
Rewards RL algorithm

COMESTOLONDON
NEW!
F I R S T S E A S O N S T A R T S
T O D A Y
The world’s first global, autonomous
racing league, open to anyone
AWS
DeepRacer

COMESTOLONDON
Build reinforcement learning model
DeepRacer League Races at AWS Summits
Winners of each DRL Race and top scorers
compete in Championship Cup at re:Invent 2019
Virtual tournaments through the year
AWS DeepRacer League
World’s first global autonomous
racing league, open to anyone

COMESTOLONDON
AWS DeepRacer
League
2018 Season
DeepRacer Speedway
MGM Grand Garden Arena

COMESTOLONDON
Dealing with documents is demanding
Manual data entry
Rules and template-based
extraction
Optical Character
Recognition (OCR)
Identify documents in any format
Extract text from those documents, accurately
Document formats are variable and change unexpectedly

COMESTOLONDON
O C R + + s e r v i c e t o e a s i l y e x t r a c t t e x t
a n d d a t a f r o m v i r t u a l l y a n y d o c u m e n t
Amazon Textract
N o M L e x p e r i e n c e r e q u i r e d
NEW!
Customer DetailsCustomer Details
Orders
Totals
Orders
Totals

COMESTOLONDON
Traditional OCR only provides a “bag of letters”
MethOd Num' C'USte'S Rand mdex ated using two types of measures. The first is the average
TM~score 8 89.7% silhouette width itself, which is a measure of the clus-
ppm 9 39,396 ter compactness and separation. In general, clustering is
305C 9 895% based on the assumption that the underlying data form
compact clusters of similar characteristics. Larger aver-
R50 7 92.096
age Silhouette Width means that the result of a clustering

COMESTOLONDON
OCR++
Amazon
Textract
Method Num. clusters Rand index
TM-score
FPFH
3DSC
RSD
VFH
Combined silhouette weights
Combined equal weights
8
9
9
7
8
7
7
89.7%
89.3%
89.5%
92.0%
85.3%
92.2%
90.2%
Aurora
Amazon Textract: an organized filing cabinet of document data

COMESTOLONDON
Amazon Textract: automatic document processing without data entry, or writing rules
Graceland, Memphis
Presley, Elvis Aaron
TCB Limited
12-12-1234
TN
01 08 1935 X
901 555-0187
3765 Elvis Presley Blvd.
38116
X
RCA Records
Rock n Roll Health
X
Presley, Elvis Aaron
Presley, Elvis AaronN A M E
Graceland, Memphis, TNA D D R E S S
12-12-1234I D
TCB LimitedC O M P A N Y
Government forms (e.g. FDA
new drug application,
financial disclosure form,
incident reporting)
Tax forms (US – e.g. W2,
1099-MISC, 990, 1040; UK –
e.g. P45; Canada – e.g. T4,
T5)
OCR++

COMESTOLONDON
Presley, Elvis AaronN A M E
Graceland, Memphis, TNA D D R E S S
12-12-1234I D
TCB LimitedC O M P A N Y
Graceland, Memphis
Presley
TCB Limited
12-12-1234
TN
901 555-0187
3765 Elvis Presley Blvd.
38116
Elvis
Elvis.Presley@yahoo.com
Amazon Textract: automatic document processing without data entry, or writing rules
OCR++

COMESTOLONDON
No master algorithm
for personalization
and recommendation
Music
recommendation
Tracks, artists, albums
Film
recommendation
Actors, directors, genres
Product
recommendation
Pricing, category, offers
Article
recommendation
Themes, geography,
breaking news

COMESTOLONDON
NEW!
R e a l - t i m e p e r s o n a l i z a t i o n a n d
r e c o m m e n d a t i o n s e r v i c e , b a s e d o n
t h e s a m e t e c h n o l o g y u s e d a t
A m a z o n . c o m
Amazon Personalize

COMESTOLONDON
Activity stream
from app
Views, signups,
conversion, etc.
Inventory
Articles, products,
videos, etc.
Demographics
(optional)
Age, location, etc.
Customized
personalization &
recommendation
API
Amazon Personalize
Amazon Personalize: machine learning personalization and recommendations

COMESTOLONDON
Activity stream
from app
Views, signups,
conversion, etc.
Inventory
Articles, products,
videos, etc.
Demographics
(optional)
LOAD DATA
(EMR Cluster)
INSPECT
DATA
IDENTIFY
FEATURES
SELECT
ALGORITHMS
SELECT
HYPERPARAMETERS
TRAIN
MODELS
OPTIMIZE
MODELS
HOST
MODELS
BUILD FEATURE
STORE
CREATE
REAL-TIME
CACHES
Customized
personalization &
recommendation
API
F u l l y m a n a g e d b y A m a z o n
P e r s o n a l i z e
Amazon Personalize
Amazon Personalize: machine learning personalization and recommendations
Age, location, etc.

COMESTOLONDON
Necessity is the mother of invention

COMESTOLONDON
The perils of
poor predictions
in forecasting
T I M
S A L E S
F O R E C A S T
S A L E S

COMESTOLONDON
T I M
F O R E C A S T
S A L E S
The perils of
poor predictions
in forecasting
S A L E S

COMESTOLONDON
NEW!
A c c u r a t e t i m e - s e r i e s f o r e c a s t i n g
s e r v i c e , b a s e d o n t h e s a m e
t e c h n o l o g y u s e d a t A m a z o n . c o m
Amazon Forecast

COMESTOLONDON
Amazon Forecast: machine learning time-series forecasting
Historical data
Supply chain,
inventory, etc.
Customized
forecasting API
Related “causal” data
Weather, special offers, product
details
Amazon Forecast

COMESTOLONDON
Historical data
Supply chain,
inventory, etc.
Customized
forecasting API
Inspect
data
Identify
features
Select
from 8
algorithms
Select
hyperparameters
Host
models
Load
data
Train
models
Optimize
models
Related “causal” data
Weather, special offers, product
details
F u l l y m a n a g e d b y A m a z o n
F o r e c a s t
Amazon Forecast
Amazon Forecast: machine learning time-series forecasting

COMESTOLONDON
Using Amazon Forecast for time-series forecasts
Any historical
time-series
Integrates with SAP and
Oracle Supply Chain
Custom forecasts
with 3 clicks
50% more
accurate
1/10th
the cost
Integrates with
Amazon Timestream
Retail demand Travel demand AWS usage
Revenue forecasts Web traffic Advertising demand
Generate forecasts for:

COMESTOLONDON
A I S e r v i c e s
Amazon
Personalize
Amazon
Forecast
V E R T I C A L
N E W N E W
V I S I O N Amazon
Textract
N E W
M L S e r v i c e s
Amazon
SageMaker
N E W
Amazon SageMaker
Ground Truth
AWS Marketplace
for ML
Amazon
SageMaker RL
Amazon
SageMaker Neo
AWS
DeepRacer
AWS
DeepRacer League
M L F r a m e w o r k s + I n f r a s t r u c t u r e
Amazon Elastic Inference
N E W
AWS Inferentia
N E W
S T O R A G E
Amazon Redshift
+ Redshift Spectrum
Amazon
QuickSight
Amazon EMR
Hadoop, Spark, Presto,
Pig, Hive…19 total
Amazon
Athena
Amazon
Kinesis
Amazon
Elasticsearch
Service
AWS Glue
A N A L Y T I C SM A C H I N E L E A R N I N G
Amazon
S3
Standard-
IA
Amazon
S3
Standard
Amazon S3
One Zone-IA
Amazon
Glacier
Amazon S3
Intelligent-
Tiering
N E W
Amazon
EBS
Amazon S3
Glacier Deep
Archive
N E W

COMESTOLONDON
M a n a g e w i t h y o u r
e x i s t i n g V M w a r e
t o o l s
S e a m l e s s l y
m i g r a t e w o r k l o a d s
R u n t h e s a m e
V M w a r e s o f t w a r e o n
A W S t h a t y o u r u n i n
y o u r d a t a c e n t e r sVMware Cloud
on AWS

COMESTOLONDON
Momentum for VMware Cloud on AWS

COMESTOLONDON
What else are
customers asking for?

COMESTOLONDON
VMware Cloud
on AWS
AWS native
AWS DESIGNED HARDWARE
The same that we run in our own data centers
OPTION 1 OPTION 2
Run AWS infrastructure on premises
for a truly consistent hybrid experience
C O M I N G I N 2 0 1 9
AWS Outposts

COMESTOLONDON
Family of hybrid options on AWS
Integrate your on-premises
resources with AWS
Use your VMware
software and tools to
run workloads on AWS
Run latency-sensitive
workloads on premises in a
way that seamlessly
integrates with AWS
Run compute on the
edge in locations with
limited connectivity
AWS Systems
Manager
VMware Cloud
on AWS
AWS Outposts
Snowball Edge
Compute-Optimized
AWS Storage
Gateway
Amazon VPC AWS Direct
Connect
AWS Identity and
Access Management
AWS Directory
Service
AWS
OpsWorks
AWS
CodeDeploy

COMESTOLONDON
Here to help you build

AWS re:Invent Comes to London 2019 - Keynote

Recommended

Recommended

More Related Content

What's hot

What's hot (20)

Similar to AWS re:Invent Comes to London 2019 - Keynote

Similar to AWS re:Invent Comes to London 2019 - Keynote (20)

More from Amazon Web Services

More from Amazon Web Services (20)

AWS re:Invent Comes to London 2019 - Keynote