Upload
others
View
8
Download
0
Embed Size (px)
Citation preview
AWS
COST OPTIMIZATION
WORKSHOP
Confidential 31.10.2018
What you’llget out of this session
Best practices on how to lower your (AWS) bill
Amore scalable, robust, dynamic architecture
More time toinnovate
Real-world customerexamples
Easy to implement
Speaker
LONG TRANRENOVACLOUD Solution Architect
HALEY PHANAWS Territory Manager - Viet Nam
QUANG TRANRENOVACLOUD COO
NGUYEN GIA HUNG AWS Solution Architect
Agenda
COST MONITORING
CHOOSE THE RIGHT SIZING
SAVING 80% with SPOT
PURCHASING OPTION
OVERVIEW OF COST OPTIMIZATION LEVERS
CASES STUDY
1OVERVIEW OF COST
OPTIMIZATION LEVERS
Confidential 31.10.2018Inc. or its Affiliates. All rights reserved. Amazon Confidential and Trademark
Overview
Impact
of cost optimisation levers
Savings
(commercial and technical)
Spot, Serverless,
Auto Scaling, and other
cost-aware architecture*
SchedulingPlan
Enterprise
Programs RIs *Typically needs upfront design investmentor re-architecture if done reactively.
Spot is a special case where sometimes
re-architecture isn’t needed.Rightsizing
pre-migration
Delete idle
resources
Rightsizingproduction
S3 intelligenttiering
ComplexityUnderpinned by cost awareness and visibility
and organisational behavioural mechanisms
Cloud Cost optimization Model
There are multiple ways of evaluating the solutions and best practices
• Infrastructure Configuration
• Cloud Management Tools
• Tuning/Optimization
• Workload Orchestration Tools
• Native Cloud Provider Optimization and Cost Management Tools
Best practices
• Reserved Instances
• Auto-Parking, Dev/Test environments
• Auto-Scaling Policies and Strategy
• Storage Optimization
Cloud Cost Optimization : A Logical Approach
Performance vs Cost Improvement
Confidential 31.10.2018
Cost optimisation activity and behaviour requires
motivation, ability, and a prompt
Motivation
+ Executive support and consistent comms.
+ External cost pressure
- Engineers feel they need to
prioritise performance / uptime / features
Ability
+ Single threaded authority for optimisation results
+ Training / tools
+ Time given to Engineers to optimise
- Org. policies that slow action
Prompts
+ Cost dashboards relevant to individuals
+ Daily/weekly cadence that includes review of spend
Promptssucceed
here
Promptsfail here
Ability
Motivation
Cost-effective
resources
Matching supply
with demand
Expenditure
awareness
Optimizing
over time
Cost optimization areas of focus
2COST MONITORING
Confidential 31.10.2018
Simple, La
nment Dyna
Pick the cost tool that provides the visibility you need
3. AWS Cost Explorer2. AWS Billing
console(and AWS Budgets)
3rd4. party tools
AWS Billing File Analysis,or DIY dashboards
1. Monthly AWS Invoice
Small, SimpleStatic environment
Large, Complex,Dynamic environment
MONTHLY INVOICES
Use theAWS Billingand Cost Management dashboard
DivedeepwithAWS CostExplorer—Now withAPI
Third Party Tools
AWS COST EXPLORER
AWS BUDGET
DEMO
3CHOOSE THE RIGHT SIZING
Cloudwatch Agent
https://amzn.to/2y9XP7Q/
Right-sizing starts with monitoring
Get startedquickly
Low utilization
© 2020, Amazon WebServices, Inc. or its affiliates. All rightsreserved.
High utilization
Opportunity: Most instances aren’t very busy
Select the cheapest instance available
• Ensure you meet performance requirements
• Consider different instance families, not just sizes
Analyze using CloudWatch metrics• Monitor CPU, RAM, storage, and network utilization
• Identify potential instances that can be downsized
• Set up custom metrics (i.e. RAM) where needed
Rule of thumb: Right-size first, then reserve
Right-sizing: Amazon EC2 Instance
AWS resourcesmetadata
Simplifying compute optimization
AWS Compute
Optimizer
Identify optimal AWS
Compute resources for
your workloadsMettle scans your AWS
infrastructure and uses
machine learning to
automatically identify
optimal AWS resources
for your workloads
Identifies workload
characteristics and
profile based on the
data gathered
Matches the resource
requirements of your
workloads to optimal
AWS resources by
recommending
Helps you visualize
what-if scenarios based
on the recommended
EC2 Instance
EC2 Auto Scaling
Groups
CloudWatch
Metrics
AWS Compute Optimizer
Recommends optimal instances forAmazon EC2andAmazon EC2Auto Scaling groups from 140+instances from M, C,R,T,andXfamilies
Applies insights from millions
of workloads to make
recommendations
Savestime comparing and
selecting optimal resources
for your workloads
Lower costsand improve workload performance
© 2020, Amazon WebServices, Inc. or its affiliates. All rightsreserved.
Easy to choose with AWS Compute Optimizer
New services that recommend optimalAWScompute resources to reducecosts up to 25%
Recommends optimal EC2 instances
Optimizes performanceand reducescosts by
making recommendations to help you right- sizecompute to your workloads
Analyzes Amazon CloudWatch metrics and considersAuto Scaling group configuration for intuitive and actionablerecommendations
Up to three recommendations per workload
Available at no additional charge
© 2020, Amazon WebServices, Inc. or its affiliates. All rightsreserved.
Capacity Optimized
• Prioritize deploying Spot Instances into greater Spot pool
capacity order to lower the chance of interruptions
Lowest cost• Prioritize cost by selecting a mix of On-Demand and Spot
Instances to launch based on the lowest available price
Right-sizing: Using Auto Scaling Group
Automatically scale instances across instance families and
purchase options in a single ASG to optimize cost
Prioritized list• Use a prioritized list for On-Demand instance types to
scale capacity during an urgent, unpredictable event to
optimize performance
AZ1 and AZ2
Spot Instances
On-Demand Instances
Reserved Instances
& Saving Plans
EC2 Auto Scaling allow you to:
• React dynamically to changes in load
• Schedule regular workloads
• Optimize your instance usage
• Reduce over-provisioning
• No cost service!
AWS Auto Scaling• Unified scaling for your
cloud applications
Right-sizing: Using Auto Scaling Group
Amazon S3 offers a range of storage classes
• Standard
• Standard – Infrequent Access
• One Zone – Infrequent Access
• Glacier
• Glacier Deep Archive
• Reduced Redundancy (no longer recommended
Keypoints to note• Standard, Standard-IA and One Zone-IA are “real-time”
storage tiers
• Glacier and Glacier Deep Archive are “near-line” storage
tiers
Right-sizing: Amazon S3
Confidential 31.10.2018
Right-sizing: Amazon S3
Rule of thumb: if you’re retrieving an object once per month or more,Standard is more cost effective storage class than Infrequent Access
• Remember, storage classes can be set on a per-object basis, not just per bucket
• Managing storage classes on a per-object basis can be complex and time-consuming
Consider using Amazon S3 Intelligent-Tiering if you have changing orunknown access patterns
• Automatically moves your data based on changing access patterns
• Moves data between Standard and Infrequent-Access tiers
• Additional management charge ($0.0025 per 1,000 objects per month)
© 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Data Transfer Costs
Region 1
Availability Zone 1 servers
using public IP
$0.01 / GB
AWS
Region 2
Availability Zone 1
Availability Zone 2Availability Zone 2 servers
using public IP
1 21
Free65
3
4
$0.02 / GB
Transfer in free
$0.01 / GB
Between AZ
Internet
Transfer out $0.09 / GB
Transfer in free
Data Transfer Cost Optimization Tips
Analyze Data Transfer Costs using Cost Explorer
Use Private IP for Internal Services Communication
Use VPC End-Points for S3 and DynamoDB
Consider using CloudFront to reduce Data Transfer Out Cost
•
Re-architecting and designing for cost
Serverless &AWS Lambda
Amazon EC2 Spot
Static Web Hosting on S3and using S3 Select
ELB toApplication Load Balancer
Deliver content withAWS CloudFront and Caching
(lower compute and data transfer)
Containerisation (staff productivityand compute utilisation)
Open source platforms & databases(reduced licensing cost)
DynamoDB on-demand+ Auto Scaling
Running resources in acheaper region and assess
level of redundancy
Low cost security monitoring viaAWS GuardDuty
QUICK WIN:
SWITCH IT OFF !
What’s the best way to turn off
resources outside of work hours?
Amazon EC2 Cost Optimisation non-prod
100.0
71.4
35.7
29.8
0
20
40
60
80
100
24 x7 24 x5 12 x5 10 x5
% RunningTime
Up to 70%
savings for non-
production
workloads
AWS Instance Scheduler
• AWS-provided solution
• Custom start & stop schedules
• Works with EC2 &RDS instances
• Deploy usingCloudFormation
• Selectively tag instances to schedule
• Multiple schedules perinstance
• 5-minute granularity
https://aws.amazon.com/answers/infrastructure-
management/instance-scheduler/
Instance Scheduler
Demo
4PURCHASING OPTION
Optimizing Amazon EC2 cost and capacity
Pricing Capacity
Capacity management made easyon the broadest
and deepest compute platform
© 2020, Amazon WebServices, Inc. or its affiliates. All rightsreserved.
Achieve optimal
Price/performance
With different
Purchase models
Amazon EC2 purchase options
On-Demand
Pay for compute capacity
by the second with no
long-term commitments
Spiky workloads to define
needs
Reserved Instances
(RIs)
Make a 1- or 3-year
commitment and receive
a significant discount on
On-Demand prices
Committed and steady-
stage usage
Savings Plans
Same great discounts as
Amazon EC2 Ris with
more flexibility
Flexible access to
compute
Spot Instances
Spare Amazon EC2
capacity at saving of up to
90% on On-Demand prices
Fault-tolerant, flexible,
stateless workloads
To optimize Amazon EC2, combine purchase options
Scale using Spot for fault-
tolerant, flexible, stateless
workloads
Use On-Demand for new or
stateful spiky workloads
Use Ris or a Saving Plans
for know, steady-state
workloads
Save up to 90% using EC2 Spot Instances
Instances
Same infrastructure as
On-Demand and RIs
Pricing
Smooth, infrequent
changes, more
predictable
Usage
Choose different instance
types, sizes and AZs in a
single fleet or EC2 Auto
Scaling Group
Capacity
Interruptions only happen
if OD needs capacity
Pricing is based on longterm supply and demand trends; no bidding!
AWS Savings Plan
A new flexible pricing model which helps you save up to 72% on EC2 and Fargate usage. Customers simply commit to a consistent amount of usage (e.g. $10/hour) over 1 or 3 years, and in exchange they will receive a lower price for that usage.
Every type of compute usage has an On Demand rate and a Savings Plans price. If a customer commits to $10/hour of compute usage, then they will get Savings Plans prices on all usage up to $10. Any usage beyond the commitment will be charged at regular
On-Demand rates.
How it works?What is it?
AWS Savings Plan
Flexible purchase option that offers savings of up to 72% on Amazon EC2 and AWS Fargate usage
Easy to use Receive lower rates automatically in
exchange for a monetary commitment
Significant SavingsSelect from two types of savings plans to
access prices of up to 66% off on Compute Savings Plans and 72% off on
EC2 Instance Plans
Flexible Make a single commitment that
applies across multiple AWS Compute Services, even as your
requirements change
Type of Savings Plan
Offer the greatest flexibility, up to 66% off (same prices as Convertible RIs)
FLEXIBLE
ACROSS
Instance family: E.g. Move from C5 to M5
Region: E.g. change from EU (Ireland) to EU (London)
OS: E.g. Windows to Linux
Tenancy: E.g. switch Dedicated tenancy to Default tenancy
Compute options: E.g. move from EC2 to Fargate
Compute Savings Plans
Provide the lowest prices, up to 72% off (same as Standard RIs) on the selected instance family (e.g. C5 or M5), in a specific AWS region
FLEXIBLE
ACROSS
Size: E.g. move from m5.xl to m5.4xl
OS: E.g. change from m5.xl Windows to m5.xl Linux
Tenancy: E.g. modify m5.xl Dedicated to m5.xl Default tenancy
EC2 Instance
Savings Plans
AWS Savings Plan
Compute Savings Plans
EC2 Instance Savings Plans
Convertible RIs* Standard RIs
Savings over On Demand Up to 66% Up to 72% Up to 66% Up to 72%
Low price in exchange for monetary commitment
Pricing automatically applies to any instance families
Pricing automatically applies to any instance size ** **
Pricing automatically applies to any Tenancy, or OS
Automatically apply to Fargate Usage
Pricing automatically applies to across any AWS Region
1- and 3-year Term length options
5Saving up to 80% using Spot
and “Spot”
Confidential - 2020
Capacity
Interruptionsonly happen if OD needs capacity
Pricing
Smooth, infrequentchanges,
more predictable
Instances
Same infrastructureas
On-Demand andRIs
Usage
Choose different instancetypes, sizes,andAZs in
asingle fleet or EC2Auto Scaling group
Pricing is based on long-term supply and demand trends; no bidding!
© 2020, Amazon WebServices, Inc. or its affiliates. All rightsreserved.
Save up to 90% using EC2 Spot Instances
Low, predictableprices
Up to 90%discount overOn-Demand prices
Faster results
Increase throughput up to 10xwhile staying in budget
Easy to use
Launch throughAWSservices (e.g.,Amazon ECS,Amazon EKS,AWS
Batch,Amazon SageMaker,Amazon EMR)or integrated third parties
Why Spot Instances?
© 2020, Amazon WebServices, Inc. or its affiliates. All rightsreserved.
Amazon Elastic
KubernetesService
(Amazon EKS)
Connectionbetween termination requests fromAWS infrastructure to nodes
Tasks running on Spot Instances will automatically be triggered for shutdown before
the instance terminates, and replacement tasks will be scheduled elsewhere on
the cluster
Amazon Elastic
ContainerService
(Amazon ECS)
Handling Spot interruptions
© 2020, Amazon WebServices, Inc. or its affiliates. All rightsreserved.
Lessthan 5%of Spot Instanceswere interrupted in the last 3months
Minimal interruptions
Check for 2-minute interruption notification via instance
metadata orAmazonCloudWatch events, and automate by
Checkpointing
Draining fromELB
Using stop-start and hibernate to restart faster
InterruptionhandlersforAmazonECSandAmazonEKS
BIG CONCERN:
INTERRUPTION and SLA ?
Giving your workloads the cloud infrastructure
they deserve: always available, always scalable,
always at the lowest possible cost
Spot Platform
Overview
on and mission critical
gle point of failure is
n.
cti
sin
io
?
du
a
uct
n
How it Works
Spotinst enables companies to reliably run pro
applications on Spot Instances. Anything without
60-80% cost reddelivered at 100% availability with
Authenticatio
User (Business)Management
SaaS•
•
•
•
Autoscaling
Provisioning
HealthChecks
Load Balancing
• Monitoring
• Orchestration
Algorithms & Machine Learning
Prediction is the Key
Spotinst uses predictive analytics algorithms and historical data to
identify and predict Spot Instances that are about to be “interrupted”.
Prior to termination, Elastigroup automatically and seamlessly transitions applications to the least
expensive “available” instances with absolutely no downtime regardless of application type or
demand.
Prediction is the Key
Elastigroup predicts Spot Instances behavior,
capacity trends, pricing and interruptions rate.
Elastigroup
Up to 90% Cost Optimization with SLA
By predicting interruptions and fluctuations
Elastigroup is able to offensively rebalance
clusters to prevent interruption.
Elastigroup | Multi-Cloud IaaS Optimization
Automate your Elastic Applications, with up to 90% cost optimization.
Optimize Costs with SLA
Reliably leverage Cloud Excess Capacity to optimize
cost and save up to 90% on compute infrastructure
across all major cloud providers.
Simplify Operations
Scale, manage and accelerate workloads without the
complexity and risk of manually managing your
infrastructure.
Ocean | Serverless Containers
Your Containers. Zero Infrastructure Management. With 90% Cost Optimization.
Container-Driven Autoscaling
Auto-detect Pod or Task infrastructure requirements
so the appropriate instance size or type will always
be available.
Simplify Operations
Deploy more without having to manage all the details
of the underlying Container infrastructure.
Eco | Continuous Reserved Capacity Management
Intelligent Reserved Instance Lifecycle Automation. With 75% Cost Optimization.
Managed RI Lifecycle
Comprehensive analysis of compute workloads, RI
buying and selling in the AWS Marketplace is
automated to ensure your workloads is running at
optimal pricing.
Finance & DevOps Synergy
Full visibility into compute consumption and
automation of optimal RI strategies.
Finance and DevOps teams can easily collaborate
on managing cloud cost.
1 yr Standard RI
3 yr Standard RI
1 yr Convertible RI
3 yr Convertible RI
40%
20%
8%
7%
13%
10%
2%
1 & 3 yr Compute Savings Plans
1 & 3 yr EC2 Savings Plans
100% Coverage
Marketplace Gold
~65% Sellable
Eco - Complete Portfolio of Commitment Plans
Comprehensive Management of Multi-Cloud Environments
Management
3B+Compute Hrs / Per Month
55+Countries Worldwide
500K+Concurrent Nodes Under
73%Average Savings
We’re on a mission to deliver scalable and reliable cloud
infrastructure at the lowest possible cost.
6Cases Studies
Saving options
for a big Hotel chain
Conclusion
• Purchase Reserve for RDS
• Consider Saving plan to save 30% computing cost
• Using SPOT with “SPOT” to saving more 70-80% computing cost
Saving review for a Start up
Quick Review for Start up
• Hiện tại AWS chỉ có 1 tài khoản cho mọi môi trường trên Cloud
• Đang chạy 3 môi trường:- Production- UAT/Staging
• Billing: Trung bình là 4000$ 1 tháng- RDS: gần 1000$- EKS: tầm 3000$ (chính)
• Hệ thống:- Hệ thống chạy 24/7 liên tục.- EC2/EKS/RDS instances hiện tại đều là On-demand.
Conclusion
• Purchase Reserve for RDS
– Consider not using multi AZ
• Consider Saving plan to save 30% computing cost
• Using SPOT with “SPOT” to saving more 70-80% computing cost
Q&A ?
Thank you
HONG KONG
9/F China Merchants Building
303-307 Des Voeux Road
Central, Sheung Wan
Hong Kong
HANOI
16th floor Ladeco Building
266 Doi Can, Ba Dinh
Vietnam
HO CHI MINH CITY
9th floor Nice Building
467 Dien Bien Phu Street
Binh Thanh, Vietnam
Mail : [email protected]
Call Us : +84 2422209657
www.renovacloud.com