12. AI Services
AI Platform
AI Engines
Amazon
Rekognition
Amazon
Polly
Amazon
Lex
More to come
in 2017
More to come
in 2017
Apache
MXNet
TensorFlow Caffe Theano KerasTorch CNTK
Amazon AI: Democratized Artificial Intelligence
Amazon
Machine Learning
EMR & Spark AWS Batch
13. New P2 GPU Instance Types
• New EC2 GPU instance type for accelerated computing
• Offers up to 16 NVIDIA K80 GPUs (8 K80 cards) in a single
instance
• The 16xlarge size provides:
• A combined 192 GB of GPU memory, 40 thousand CUDA cores
• 70 teraflops of single precision floating point performance
• Over 23 teraflops of double precision floating point performance
• Example workloads include:
• Deep learning, computational fluid dynamics, computational finance,
seismic analysis, molecular modeling, genomics, VR content
rendering, accelerated databases
14. P2 Instance Types
Three P2 instance sizes:
Instance
Size
GPUs GPU
Peer to
Peer
vCPU
s
Memory
(GiB)
Network
Bandwidth
*
p2.xlarge 1 - 4 61 1.25Gbps
p2.8xlarge 8 Y 32 488 10Gbps
p2.16xlarge 16 Y 64 732 20Gbps
*In a placement group
15. Apache MXNet
Programmable Portable High Performance
Near linear scaling
across hundreds of GPUs
Highly efficient
models for mobile
and IoT
Simple syntax,
multiple languages
Most Open Best On AWS
Optimized for
deep learning on
AWS
Accepted into the
Apache Incubator
17. One-Click GPU or CPU
Deep Learning
AWS Deep Learning AMI
Up to~40k CUDA cores
Apache MXNet
TensorFlow
Theano
Caffe
Torch
Keras
Pre-configured CUDA drivers, MKL
Anaconda, Python3
Ubuntu and Amazon Linux
+ AWS CloudFormation template
+ Container image
18. Application Examples | Jupyter Notebooks
• https://github.com/dmlc/mxnet-notebooks
• Basic concepts
• NDArray - multi-dimensional array computation
• Symbol - symbolic expression for neural networks
• Module - neural network training and inference
• Applications
• MNIST: recognize handwritten digits
• Check out the distributed training results
• Predict with pre-trained models
• LSTMs for sequence learning
• Recommender systems
• Train a state of the art Computer Vision model (CNN)
• Lots more..
20. What is serverless computing?
• VMs
• Machine as the unit of scale
• Abstracts the hardware
• Containers
• Application as the unit of scale
• Abstracts the OS
• Serverless
• Functions as the unit of scale
• Abstracts the language runtime
Amazon ECS
Amazon EC2
AWS Lambda
22. AWS Lambda Programming Model
Bring your own code
• Node.js, Java, Python, .NET
• Bring your own libraries
(even native ones)
Simple resource model
• Select power rating from
128 MB to 1.5 GB
• CPU and network allocated
proportionately
• Reports actual usage
Programming model
• AWS SDK built in (Python
and Node.js)
• Lambda is the “webserver”
• Use processes, threads,
/tmp, sockets normally
Stateless
• Persist data using Amazon
DynamoDB, S3, or Amazon
ElastiCache
• No affinity to infrastructure
(can’t “log in to the box”)
25. “I want to sequence functions”
“I want to select functions based on data”
“I want to retry functions”
“I want try/catch/finally”
Functions into apps
“I have code that runs for hours”
“I want to run functions in parallel”
28. Application Lifecycle in AWS Step Functions
Visualize in the
Console
Define in JSON Monitor
Executions
29. Integrate with other AWS services
• Create state machines and Activities with AWS
CloudFormation
• Call AWS Step Functions with Amazon API Gateway
• Start state machines in response to events, or on a
schedule, with Amazon CloudWatch Events
• Monitor state machine executions with Amazon CloudWatch
• Log API calls with AWS CloudTrail
32. AWS X-Ray – GA and Available for Lambda
Analyze and debug production, distributed
applications
Simple setup: instrument your application with X-
Ray SDK and install X-Ray Daemon
End-to-End Tracing, cross-service view of
requests made to your application.
33. AWS X-Ray
Service Map provides a view of connections
between services in your application and aggregated
data for each service, including average latency and
failure rates.
Data Annotation and Filtering
39. • Integration
tests with
other systems
• Load testing
• UI tests
• Penetration
testing
Release processes have four major phases
Source Build Test Production
• Check-in
source code
such as .java
files.
• Peer review
new code
• Compile code
• Unit tests
• Style checkers
• Code metrics
• Create
container
images
• Deployment
to production
environments
51. Amazon Lightsail - Everything you need to
launch a simple website or web app
Lightsail virtual
private server
SSD-based storage Networking & data
transfer
DNS management Static IP Access to AWS
services
Monash University is using Amazon Lightsail to rapidly re-platform a number of CMS services in a simple and
cost-effective manner. They have already migrated 50 workloads and are now thinking of creating an internal
CMS service based on Lightsail to allow staff and students to create their own CMS instances in a self-service
manner.
56. • A fully managed file system for Amazon EC2 instances
• A file system interface with file system access semantics that
works with standard operating system APIs
• Sharable across thousands of instances
• Designed to grow elastically to petabyte scale
• Built for performance across a wide variety of workloads
• Highly available and durable
• Strong consistency
Amazon Elastic File System (EFS)
57. We are growing by leaps and bounds, and our core offering
is all about better support delivery. During the course of
developing our next-generation internal support system,
we never wanted to worry about scale again, yet we had
existing architectural commitments that meant a
distributed file solution was required. We chose Amazon
EFS because it was the only option available that scaled
both capacity and performance – without the up-front
payments or the management overhead of traditional
models. Now we can just focus on supporting our
customers to help them support their customers.
- Sam Caldwell, Sr. Systems Engineer
”
“
Customer Example
60. The Amazon CloudFront Service
Global Content Delivery Network with Massive Capacity and Scale
Optimized for Performance and Scale
Built in Security Features
Self-Service Full Control Configurations
Robust Real Time Reporting
Amazon
CloudFront
Static and Dynamic Object and Video Delivery