26 SEP,2024

© 2024, Amazon Web Services, Inc. or its affiliates. All rights reserved.
 Compute best practices and design patterns
for resiliency

Sreekumar Nair Krishnaji Panse

Senior Solutions Architect Senior Solutions Architect
AWS AWS

© 2024, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2024, Amazon Web Services, Inc. or its affiliates. All rights reserved.
 Applications power businesses of all kinds

© 2024, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2024, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Organizations are moving applications and data to the
cloud
Goals

Accelerate business transformation

Increase agility and innovate quickly

Reduce operating costs

Legacy Access cloud scale, performance, and operations Modern

Improve security and compliance

© 2024, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2024, Amazon Web Services, Inc. or its affiliates. All rights reserved.
 Compute platform
for every workload

© 2024, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2024, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Provide customers with tools and
services to securely and reliably
run virtually any workload

© 2024, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2024, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Amazon Elastic Compute Cloud (Amazon EC2) in 2006

One size fits all Pay for what you use Scale up or down as
required quickly

© 2024, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2024, Amazon Web Services, Inc. or its affiliates. All rights reserved.
 Broadest and deepest platform choice
Categories Capabilities Options

Choice of processor
(AWS, Intel, AMD)
General purpose Fast processors
(up to 4.5 GHz)

750+
Burstable
High footprint
Compute-intensive (up to 24 memory TiB)
Amazon Elastic Block Store
Instance storage (Amazon EBS)
Memory-intensive (HDD and NVMe)
Amazon Elastic Inference
Storage (high I/O) Accelerated computing
(GPUs and FPGA) Elastic Fabric Adapter (EFA) for virtually every
Dense storage workload and
Networking business need
(up to 100 Gbps)
GPU compute
Bare metal
Graphics-intensive
Size
(nano to 32xlarge)

© 2024, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2024, Amazon Web Services, Inc. or its affiliates. All rights reserved.
 Amazon EC2 instance characteristics

Instance Additional
CPU generation capabilities

Memory
M5ad.xlarge
Storage Instance Processor Instance size
family type

Network
Instance type

© 2024, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2024, Amazon Web Services, Inc. or its affiliates. All rights reserved.
 Broadest choice of processors and architectures

Right compute for the workload

Intel Xeon AMD EPYC AWS Graviton Apple M1

Scalable processors processors processors processors

© 2024, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2024, Amazon Web Services, Inc. or its affiliates. All rights reserved.
 Broadest choice of processors and architectures

Right compute for the workload

Intel Xeon
Scalable processors

© 2024, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2024, Amazon Web Services, Inc. or its affiliates. All rights reserved.
 Innovating with Intel
17 years of collaboration and innovation with AWS

Collaboration Integration Speed

Deep engineering collaboration Over 400 Amazon EC2 instances Fastest processor in the cloud and
across the AWS portfolio are powered by Intel processors widest selection of Ice Lake instances

Recent Intel-based instances

I4I M6I(D)(N) C6I(D)(N) C7I-flex R6I(D)(N) R7I R7Iz M7I M7I-flex

Storage- General Compute- Cost-optimized Memory- Memory- Memory- General Cost-optimized
optimized purpose optimized general purpose optimized optimized optimized purpose general purpose

© 2024, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2024, Amazon Web Services, Inc. or its affiliates. All rights reserved.
 Broadest choice of processors and architectures

Right compute for the workload

AMD EPYC
processors

© 2024, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2024, Amazon Web Services, Inc. or its affiliates. All rights reserved.
 Innovating with AMD
10% lower cost versus comparable x86 instances

Flexibility Better economics Seamless migration

Help you optimize both cost and Deliver up to 10% lower cost Easily migrate applications to the
performance for your workloads versus comparable instances new AMD-based variants with little
to no modification

AMD-based instances

G4DN G4AD HPC7A M7A C7A R7A

Machine learning GPU-optimized HPC-optimized General purpose Compute-optimized Memory-optimized

© 2024, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2024, Amazon Web Services, Inc. or its affiliates. All rights reserved.
 Broadest choice of processors and architectures

Right compute for the workload

AWS Graviton
processors

© 2024, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2024, Amazon Web Services, Inc. or its affiliates. All rights reserved.
 Innovating with AWS Graviton2
40% better price performance for a broad range of workloads

Best price performance Extensive ecosystem Enhanced security

Delivers up to 40% better price Supported by popular Linux OS Provides key capabilities for
performance over comparable along with popular applications and application security, including
x86-based instances services from AWS and ISVs 256-bit DRAM encryption

AWS Graviton2-based instances

C6G(D) M6G(D) R6G(D) C6GN X2GD T4G IM4GN IS4GEN I4G G5G
Compute- General Memory- Network- Enhanced Burstable Storage- Storage- Storage-
optimized purpose optimized optimized memory optimized optimized GPUs
optimized

© 2024, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2024, Amazon Web Services, Inc. or its affiliates. All rights reserved.
 Innovating with AWS Graviton3
Provides a 25% performance improvement compared to AWS Graviton2

Improved Faster Energy

performance memory efficient

Up to 25% higher compute Supports DDR5 memory to Consumes 60% less power
performance and 2x higher provide 50% more memory for the same performance
floating point to accelerate bandwidth over DDR4 memory compared to other CPUs
compute-intensive workloads

AWS Graviton3-based instances

C7G(D) M7G(D) R7G(D) C7GN HPC7G

Compute- General Memory- Enhanced Enhanced vector
optimized purpose optimized networking performance (Graviton3E)

© 2024, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2024, Amazon Web Services, Inc. or its affiliates. All rights reserved.
AWS Graviton3 uses up to
60% less energy to compute
the same workload as
comparable x86 processors

© 2024, Amazon Web Services, Inc. or its affiliates. All rights reserved. Source: https://aws.amazon.com/ec2/graviton/graviton-sustainability/
© 2024, Amazon Web Services, Inc. or its affiliates. All rights reserved.
 Innovating with AWS Graviton4
Best price performance for memory-intensive workloads on Amazon EC2

Improved performance Large instance size Enhanced security

Up to 30% better performance Up to 3x more vCPUs and Fully encrypting all high-speed
compared to Graviton3-based memory than the seventh- physical hardware interfaces
R7g instances generation AWS Graviton3-
based R7g instances

AWS Graviton4-based instances

R8G
Memory-optimized

© 2024, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2024, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Beyond Amazon EC2: AWS Managed Services
(AMS) supporting AWS Graviton
Extending the AWS Graviton price performance to managed services

Databases

Amazon DocumentDB Amazon Aurora Amazon Relational Amazon Amazon Amazon

Database Service ElastiCache MemoryDB Neptune
(Amazon RDS)

Analytics Compute Machine learning

Amazon Amazon EMR AWS Lambda AWS Fargate AWS Elastic Amazon
OpenSearch Beanstalk SageMaker
Service

© 2024, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2024, Amazon Web Services, Inc. or its affiliates. All rights reserved.
 Broadest choice of processors and architectures

Right compute for the workload

Apple M1
processors

© 2024, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2024, Amazon Web Services, Inc. or its affiliates. All rights reserved.
 Amazon EC2 Mac instances
On-demand Apple silicon macOS environment for the first time on AWS

Powered by Improved Harness

Apple silicon performance the cloud

Apple M1 chip integrates the CPU, Up to 4x better build performance Provision macOS environments within
GPU, neural engine, I/O, and much compared to on-premises and up to minutes and only pay for what you use;
more into a single tiny chip 60% better price performance offload the heavy lifting that comes with
compared to x86 Mac instances managing infrastructure to AWS

© 2024, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2024, Amazon Web Services, Inc. or its affiliates. All rights reserved.
More and more companies are
deploying machine learning to
improve their customer experience

© 2024, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2024, Amazon Web Services, Inc. or its affiliates. All rights reserved.
 Supporting ML workloads
Largest selection of ML instances in the cloud

Training
Machine learning

P4D P4DE P5 DL1 TRN1

NVIDIA A100 NVIDIA A100 NVIDIA H100 Habana Gaudi AWS Trainium
tensor core tensor core tensor core Intel
Inference

Platform
TensorFlow

INF2
MXNet
G6 G5G INF1
NVIDIA L4T NVIDIA T4G AWS Inferentia AWS Inferentia2
tensor core tensor core PyTorch

© 2024, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2024, Amazon Web Services, Inc. or its affiliates. All rights reserved.
 Amazon EC2 Auto Scaling
Add or remove capacity to meet changing demand

AZ1 AZ2 AZ3

Application Amazon EC2 Savings Plans

Spot Instances

Lifecycle Fault tolerant Predictive

management application scaling
AWS Graviton

Improve fault tolerance Improve application Lower cost

availability
Instance lifecycle management Match capacity to demand, Scale across instance types,
across Availability Zones proactive provisioning purchase options, and architectures
to reduce over-provisioning

© 2024, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2024, Amazon Web Services, Inc. or its affiliates. All rights reserved.
 Attribute-based instance type selection
Provision capacity based on instance attributes according to compute requirements

Benefits: Configuration details:

▪ Price protected ▪ ABIS requires the use of launch
▪ Instance flexibility templates
▪ Simplifies provisioning ▪ Supports only “lowest-price”
▪ Mixed instance groups On-Demand allocation strategy
▪ Future proof ▪ Supports “price-capacity-
▪ AWS Management Console, CLI, and optimized”, “capacity-
SDKs
optimized”, and “lowest-price”
▪ Preview instances
spot allocation

© 2024, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2024, Amazon Web Services, Inc. or its affiliates. All rights reserved.
 Cost-optimization best practices
Cost reduction strategies for Amazon EC2

Diversify your Amazon EC2 Choose the right purchase Match capacity to demand
instance types models
AWS Graviton–based instances offer up Savings Plans offer a flexible pricing Available tools like AWS Compute
to 40% better price performance, and model with savings of up to 72% on Optimizer and AWS Cost Explorer
AMD-based instances deliver a 10% your AWS compute usage compared to provide easy to implement right-
sizing recommendations for your
savings versus comparable On-Demand
workloads
x86-based instances

© 2024, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2024, Amazon Web Services, Inc. or its affiliates. All rights reserved.
 Amazon EC2 purchase options
Optimize compute purchase strategies

On-Demand Instances Savings Plans Amazon EC2 Spot Instances

Pay for compute capacity by the Same great discounts as Amazon Spare Amazon EC2 capacity at savings
second with no long-term EC2 Reserved Instances (RI) with of up to 90% on On-Demand prices
commitments more flexibility

Ideal for spiky workloads as Flexible access Fault-tolerant, flexible,

you begin defining your to compute and stateless workloads
needs

© 2024, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2024, Amazon Web Services, Inc. or its affiliates. All rights reserved.
 Combining purchase options
Optimize compute purchasing strategies

Scale using Spot Instances

for fault-tolerant, flexible,
stateless workloads

On-Demand Instances, for

new or stateful spiky
workloads

Use Savings Plans for

known, steady-state
workloads

© 2024, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2024, Amazon Web Services, Inc. or its affiliates. All rights reserved.
 AWS Compute Optimizer
Get recommendations to optimize use of AWS resources

Recommendation preference Resource analysis Reconfigure resources

Lowers costs and improves Applies insights from millions of Saves time comparing and selecting
workload performance workloads to make optimal resources for your workload
recommendations

Recommends optimal instances for AWS Lambda, Amazon EC2, and Amazon EC2 Auto Scaling groups from
over 145 instances in the M, C, R, T, and X families

© 2024, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2024, Amazon Web Services, Inc. or its affiliates. All rights reserved.
 Resilience
Ability of a workload to recover from infrastructure or service disruptions

High availability (HA) Disaster recovery

Resistance to common failures through design Returning to normal operation at a recovery site,
and operational mechanisms at a primary site within specific targets, for failures beyond HA
The mental model

capabilities

Core services and design goals to Backup and recovery, data bunkering, and
meet availability goals managed recovery objectives

Continuous improvement
CI/CD, observability, and moving beyond pre-deployment
testing towards chaos engineering patterns

© 2024, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2024, Amazon Web Services, Inc. or its affiliates. All rights reserved.
 Multi-AZ for high availability (HA)

Availability Zone A Availability Zone B Availability Zone C

Internet gateway
VPC
Amazon Simple
Storage Service
(Amazon S3)
NAT gateway Elastic Load NAT gateway NAT gateway
Balancing

Amazon DynamoDB
Compute Amazon Auto Scaling group Compute Amazon Compute Amazon
instance EBS instance EBS instance EBS

Amazon RDS Amazon RDS Amazon RDS

(primary) (standby) (read replica)

© 2024, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2024, Amazon Web Services, Inc. or its affiliates. All rights reserved.
 Multi-Region for disaster recovery (DR)

Each AWS Region has multiple AZs

Region Region

Transit AZ
Backup

Snapshot
AZ
Active Passive
(standby)

Transit AZ

Region Region

A Region is a physical location

Backup
in the world

Snapshot
Active Active

© 2024, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2024, Amazon Web Services, Inc. or its affiliates. All rights reserved.
What is the Modernization Pathway Recommender?
Under beta program

Modernization Pathway
Recommender

1. Fully automated workflow plug-in leveraging AWS Systems Manager Automation.

2. Securely access software inventory data from Amazon EC2 instances (and virtual machines) stored in Amazon
S3, and gather application metadata.
3. Process application metadata by invoking API calls on your behalf.
4. Generate suitable modernization pathways leveraging generative AI foundation models (Amazon Bedrock).
5. Optionally tag virtual machines with application roles to provide application context to recommendations.
6. Suggest a brief modernized target architecture, with resiliency architecture best practices.

© 2024, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2024, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Visit the Migrate. Modernize. Build. resource hub
Dive deeper into these resources:

• 6 steps to success with generative AI

• Microsoft workloads on AWS: 12 stories of migration and modernization
• 5 ways a secure cloud infrastructure drives innovation
• 10 ways to optimize costs and innovate with AWS
• Containers and serverless recommendation guide https://tinyurl.com/migrate-modernize-build

• Running Windows workloads on AWS: Your questions answered

• Top 10 reasons to choose AWS for SAP

… and more!

Visit resource hub

© 2024, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2024, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Thank you for attending AWS Innovate – Migrate. Modernize. Build.

We hope you found it interesting! A kind reminder to complete the survey.

Let us know what you thought of today’s event and how we can improve the event
experience for you in the future.

[email protected]

twitter.com/AWSCloud

facebook.com/AmazonWebServices

youtube.com/user/AmazonWebServices

linkedin.com/company/amazon-web-services

twitch.tv/aws

© 2024, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2024, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Thank you!

© 2024, Amazon Web Services, Inc. or its affiliates. All rights reserved.
© 2024, Amazon Web Services, Inc. or its affiliates. All rights reserved.