SlideShare a Scribd company logo
RightScale Webinar: Decoding AWS Reserved Instances (RIs) What It Means for Cloud Pricing
•  Hassan Hosseini
•  Product Manager, Cloud Analytics
•  Kim Weins
•  VP Marketing
Speakers
•  Understanding the New AWS Reserved Instances
•  Comparing AWS to Google
•  Breakeven Analysis
•  Factoring in Future Price Drops
Agenda
What Happened
•  Quick recap: What is a Reserved Instance
•  A commitment from you to AWS to use a resource for a given period of time
•  For that commitment, you get a discounted price (between 20% - 75%)
•  You also get a capacity reservation
•  It is not a VM. It is not a dedicated machine.
•  What changed:
•  Light, Medium and Heavy RIs were discontinued (you could see it coming)
•  All Upfront, Partial Upfront and No Upfront RIs were introduced – discounts based on timing of payment
•  All the new RIs are billed 24 hours a day – regardless of whether they are used
•  What hasn’t changed:
•  You still have to commit to an instance type, platform (OS), and Availability Zone
•  You can request modifications to move AZs or change instance sizes within a family
•  Terms are 1 and 3 years
•  Your current RIs
RI Savings Vary
5
RI price points and discounts are different
for every region, instance type on OS!!!
No Upfront
 Partial Upfront
 All Upfront

Terms Available

1 year
 1 year or 3 year
 1 year or 3 year

Savings for 1-Year RIs 
(Linux)

20-50%
 25-58%
 27-58%

Savings for 3-Year RIs
(Linux)

N/A
 41-74%
 45-75%

Hours Billed

24 hrs a day
 24 hrs a day
 24 hrs a day
Effective Hourly Rate for RIs
On!
Demand!
All!
Upfront!
$$$!$!
Partial!
Upfront!
!
Medium! No!
Upfront!
Effective hourly rates of 1-year RIs
Linux in US-East
Assumes 100% utilization
________!
Heavy!
m3.medium! $.0700!$.0500!
Light!
$.0433!$.0425! $.0477! $.0766!
39%
savings
38%
savings
29%
savings
RI price points and discounts are different
for every region, instance type on OS!!!
•  You are now committing to the minimum number of instances used.
•  You are charged hourly rate 24x7 regardless if you are using it
•  The Effective Hourly Rate and the discounts published by AWS are based on 24x7 usage
•  If you scale up and down, or turn off instances during weekends and holidays, you need to look closely at
breakeven points.
•  Example: There are 251 working days in 2014. If you turn off an instance on weekends or after hours, No
Upfront is not worth it.
•  Combine historic analysis with your knowledge of the future
•  Can you commit to an instance type in a specific Availability Zone?
•  Is you application’s architecture changing? (Dev/Ops team can help here)
•  Instance type consolidation is more important
•  Have a planned approach across your company on instance types and sizes
•  Consolidation will make it easier to buy and use Ris to get highest discount and minimize waster
•  Analysis on underutilized RIs (waste) becomes much more important
Purchasing New AWS RIs
7
•  Analyze past usage trends with Cloud Analytics
What and How Many to Buy
8
•  Forecasts future usage with Cloud Analytics
What-if Scenarios
9
•  How they are applied can be different to how you show/charge back
•  Here is how they are applied:
•  In Consolidated Billing, the purchasing account gets priority
•  If you have bought the RI at the payer level, then:
•  The oldest instance gets the reservation till it stops using it, then it goes to the next oldest
•  The cheapest price is applied from all your reservations
•  Here is how you can show/charge back
•  Un-Blended rate: Show the costs as they are applied (instance 1 in account 1 used it for 3
hours, instance 1 in account 2 used it for 5 hours etc). Deal with the upfront cost too.
•  Blended rate: Regardless of who used RIs or On-Demand instances, split sum based on
usage.

How RIs are Applied by AWS
10
•  Volume Discounts
•  $500K and above, you will get 5% discount
•  $4M and above, you will get 10% discount
•  These are within a region only (e.g. US-East). Discounts do not include Windows with SQL Server.
•  Spot
•  AWS selling their unused servers on an hourly bid price.
•  If your application can handle an instance suddenly disappearing with no warning, this is an option.
•  RI Marketplace
•  Sell your unwanted Reserved Instances (US only).
•  You set the upfront fee. This is what you will get if sold (minus 12% service charge).
•  Cannot sell RIs that you have got volume discounts on.
•  How long will it take to sell on the marketplace? 
•  Median of 14-19 days according to Jan 2014 analysis by http://blog.supplyframe.com
•  RightScale customer – sold all RIs (primarily Heavy) within 2-4 weeks at residual value
•  Buy shorter-term RIs
Other Ways to Save on AWS Pricing
11
AWS VS GOOGLE PRICING
12
About Google Sustained-Use Discounts
•  No commitment. The more you use an instance family during
the month, the higher the discount.
•  Hourly rate tiers down as you hit 25%, 50% and 75% of the
time in a month.
Usage Level
% of Billing Cycle
Incremental Rate 
% of On-Demand Baseline
Sample Rate
n1-standard-1
Total Cost
0-25%
 100%
 $0.07
 $13.02
25-50%
 80%
 $0.056
 $10.42
50-75%
 60%
 $0.042
 $7.81
75-100%
 40%
 $0.028
 $5.21
Monthly Cost
at 100% usage
$36.46
Google SUD uses “Inferred Instances”
14
Source: Google!
•  AWS No Upfront RIs vs. Google 100% SUD is closest to apples-to-apples

•  Comparisons based on:
•  US-East Region
•  Linux operating system
•  Assumes instances are running 24x7
•  No volume discount tiers

•  Keep in mind that this is a price comparison only. The specific performance
of each instance might vary.
•  AWS high-memory and high-CPU instances have more memory along with an SSD, that
balances Google’s lower price. Depending on workload, the high-CPU c3 series from AWS
could provide a better price/performance option.
AWS vs Google
15
16
Google
Instance Type
CPU
Cores
RAM
AWS
Instance Type
CPU
Cores
RAM
Google New
On-Demand 
(per hour)
AWS
On-Demand
(per hour)
AWS Price vs.
Google
n1-standard-1
 1
 3.75
 m3.medium
 1
 3.75
 $ 0.063
 $ 0.070
 +11.11%
n1-standard-2
 2
 7.5
 m3.large
 2
 7.5
 $ 0.126
 $ 0.140
 +11.11%
n1-standard-4
 4
 15
 m3.xlarge
 4
 15
 $ 0.252
 $ 0.280
 +11.11%
n1-standard-8
 8
 30
 m3.2xlarge
 8
 30
 $ 0.504
 $ 0.560
 +11.11%
n1-highmem-2
 2
 13
 r3.xlarge
 2
 15
 $ 0.148
 $ 0.175
 +18.24%
n1-highmem-4
 4
 26
 r3.2xlarge 
 4
 30.5
 $ 0.296
 $ 0.350
 +18.24%
n1-highmem-8
 8
 52
 r3.4xlarge
 8
 61
 $ 0.592
 $ 0.700
 +18.24%
n1-highcpu-2
 2
 1.8
 c3.large
 2
 3.75
 $ 0.080
 $ 0.105
 +31.25%
n1-highcpu-4
 4
 3.6
 c3.xlarge
 4
 7.5
 $ 0.160
 $ 0.210
 +31.25%
n1-highcpu-8
 8
 7.2
 c3.2xlarge
 8
 15
 $ 0.320
 $ 0.420
 +31.25%
n1-highcpu-16
 16
 14.4
 c3.4xlarge
 16
 30
 $ 0.640
 $ 0.840
 +31.25%
Google vs. AWS On-Demand Pricing!
Source: RightScale
After GCE price cuts of October 1, 2014
17
Google
Instance Type
CPU
Cores
RAM
AWS 
Instance Type
CPU
Cores
RAM
Google New
100% Usage
( per year)
AWS 1-year
No Upfront RI
(per year)
New AWS Price
vs. Google
n1-standard-1
 1
 3.75
 m3.medium
 1
 3.75
 $394
 $438
 +11.11%
n1-standard-2
 2
 7.5
 m3.large
 2
 7.5
 $780
 $876
 +12.36%
n1-standard-4
 4
 15
 m3.xlarge
 4
 15
 $1,551
 $1,752
 +12.99%
n1-standard-8
 8
 30
 m3.2xlarge
 8
 30
 $3,094
 $3,504
 +13.25%
n1-highmem-2
 2
 13
 r3.xlarge
 2
 15
 $911
 $964
 +5.77%
n1-highmem-4
 4
 26
 r3.2xlarge 
 4
 30.5
 $1,822
 $1,927
 +5.77%
n1-highmem-8
 8
 52
 r3.4xlarge
 8
 61
 $3,635
 $3,854
 +6.02%
n1-highcpu-2
 2
 1.8
 c3.large
 2
 3.75
 $491
 $639
 +30.36%
n1-highcpu-4
 4
 3.6
 c3.xlarge
 4
 7.5
 $981
 $1,279
 +30.36%
n1-highcpu-8
 8
 7.2
 c3.2xlarge
 8
 15
 $1,962
 $2,558
 +30.36%
n1-highcpu-16
 16
 14.4
 c3.4xlarge
 16
 30
 $3,924
 $5,116
 +30.36%
Google Sustained-Use vs. AWS 1-year No Upfront Reserved Instance Pricing!
Source: RightScale
After AWS RI model change of December 2, 2014
18
Google
Instance Type
CPU
Cores
RAM
AWS 
Instance Type
CPU
Cores
RAM
Google New
100% Usage
( per year)
AWS 1-year
Partial Upfront
RI
(per year)
New AWS Price
vs. Google
n1-standard-1
 1
 3.75
 m3.medium
 1
 3.75
 $394
 $379
 -3.78%
n1-standard-2
 2
 7.5
 m3.large
 2
 7.5
 $780
 $767
 -1.57%
n1-standard-4
 4
 15
 m3.xlarge
 4
 15
 $1,551
 $1,534
 -1.07%
n1-standard-8
 8
 30
 m3.2xlarge
 8
 30
 $3,094
 $3,051
 -1.33%
n1-highmem-2
 2
 13
 r3.xlarge
 2
 15
 $911
 $830
 -8.85%
n1-highmem-4
 4
 26
 r3.2xlarge 
 4
 30.5
 $1,822
 $1,660
 -8.89%
n1-highmem-8
 8
 52
 r3.4xlarge
 8
 61
 $3,635
 $3,320
 -8.67%
n1-highcpu-2
 2
 1.8
 c3.large
 2
 3.75
 $491
 $554
 +12.86%
n1-highcpu-4
 4
 3.6
 c3.xlarge
 4
 7.5
 $981
 $1,116
 +13.75%
n1-highcpu-8
 8
 7.2
 c3.2xlarge
 8
 15
 $1,962
 $2,215
 +12.90%
n1-highcpu-16
 16
 14.4
 c3.4xlarge
 16
 30
 $3,924
 $4,439
 +13.10%
Google Sustained-Use vs. AWS 1-year Partial Upfront Reserved Instance Pricing!
Source: RightScale
After AWS RI model change of December 2, 2014
19
Google
Instance Type
CPU
Cores
RAM
AWS 
Instance Type
CPU
Cores
RAM
Google New
100% Usage
( per year)
AWS 1-year
Partial Upfront
RI
(per year)
New AWS Price
vs. Google
n1-standard-1
 1
 3.75
 m3.medium
 1
 3.75
 $394
 $372
 -5.56%
n1-standard-2
 2
 7.5
 m3.large
 2
 7.5
 $780
 $751
 -3.71%
n1-standard-4
 4
 15
 m3.xlarge
 4
 15
 $1,551
 $1,503
 -3.05%
n1-standard-8
 8
 30
 m3.2xlarge
 8
 30
 $3,094
 $2.989
 -3.34%
n1-highmem-2
 2
 13
 r3.xlarge
 2
 15
 $911
 $813
 -10.77%
n1-highmem-4
 4
 26
 r3.2xlarge 
 4
 30.5
 $1,822
 $1,626
 -10.77%
n1-highmem-8
 8
 52
 r3.4xlarge
 8
 61
 $3,635
 $3,253
 -10.53%
n1-highcpu-2
 2
 1.8
 c3.large
 2
 3.75
 $491
 $542
 +10.54%
n1-highcpu-4
 4
 3.6
 c3.xlarge
 4
 7.5
 $981
 $1,093
 +11.43%
n1-highcpu-8
 8
 7.2
 c3.2xlarge
 8
 15
 $1,962
 $2,170
 +10.58%
n1-highcpu-16
 16
 14.4
 c3.4xlarge
 16
 30
 $3,924
 $4,350
 +10.85%
Google Sustained-Use vs. AWS 1-year All Upfront Reserved Instance Pricing!
Source: RightScale
After AWS RI model change of December 2, 2014
20
Google
Instance Type
CPU
Cores
RAM
AWS 
Instance Type
CPU
Cores
RAM
Google New
100% Usage
( per year)
AWS 3-year
Partial Upfront
RI
(per year)
New AWS Price
vs. Google
n1-standard-1
 1
 3.75
 m3.medium
 1
 3.75
 $394
 $244
 -38.22%
n1-standard-2
 2
 7.5
 m3.large
 2
 7.5
 $780
 $487
 -37.52%
n1-standard-4
 4
 15
 m3.xlarge
 4
 15
 $1,551
 $974
 -37.18%
n1-standard-8
 8
 30
 m3.2xlarge
 8
 30
 $3,094
 $1,948
 -37.00%
n1-highmem-2
 2
 13
 r3.xlarge
 2
 15
 $911
 $572
 -37.21%
n1-highmem-4
 4
 26
 r3.2xlarge 
 4
 30.5
 $1,822
 $1,144
 -37.21%
n1-highmem-8
 8
 52
 r3.4xlarge
 8
 61
 $3,635
 $2,288
 -37.06%
n1-highcpu-2
 2
 1.8
 c3.large
 2
 3.75
 $491
 $362
 -26.25%
n1-highcpu-4
 4
 3.6
 c3.xlarge
 4
 7.5
 $981
 $733
 -25.27%
n1-highcpu-8
 8
 7.2
 c3.2xlarge
 8
 15
 $1,962
 $1,466
 -25.31%
n1-highcpu-16
 16
 14.4
 c3.4xlarge
 16
 30
 $3,924
 $2,931
 -25.31%
Google Sustained-Use vs. AWS 3-year Partial Upfront Reserved Instance Pricing!
Source: RightScale
After AWS RI model change of December 2, 2014
Things to Remember
•  For AWS:
•  High-CPU and high-mem have more memory and SSD
•  You can get volume discounts (5% starting at $500K annual RI spend)
•  You can choose Spot Pricing to save more
•  You can use RI Marketplace to buy/sell RIs
•  For Google:
•  No commitment
•  No lock-in to instance types or regions
•  It’s automatic - no RI purchase decisions
•  Per-minute pricing (minimum of 10 minutes)
•  When on-demand price goes down, you get the benefit.
•  They’ve lowered On-Demand prices 38% so far in 2014
BREAKEVEN ANALYSIS
22
Calculating Breakeven for AWS RIs
•  Breakeven Defined
•  How long do I need to use an RI for it to be cheaper than On-demand?
•  This is not a cash-flow analysis, but a cost analysis
•  The Impact Upfront Commitment
•  For all the new-style AWS Ris, you are committing to the full price of the RI upfront – the
only difference is when you actually pay
•  Unlike old Light and Medium RIs, with new AWS Ris you can’t stop paying the hourly fee
even if you stop using the RI – no walkaway option
$- 
$100.00 
$200.00 
$300.00 
$400.00 
$500.00 
$600.00 
$700.00 
Month
1
Month
2
Month
3
Month
4
Month
5
Month
6
Month
7
Month
8
Month
9
Month
10
Month
11
Month
12
On-Demand
1yr No Upfront RI = 261 days to breakeven (72%)
1yr Partial Upfront RI = 226 days to breakeven (62%)
1yr All Upfront RI = 222 days to breakeven (61%)
Breakeven Analysis for m3.medium (US-East, Linux) for NEW 1-year RIs !
Source: RightScale
After AWS RI model change of December 2, 2014
$- 
$200.00 
$400.00 
$600.00 
$800.00 
$1,000.00 
$1,200.00 
$1,400.00 
$1,600.00 
$1,800.00 
$2,000.00 
Month1
Month2
Month3
Month4
Month5
Month6
Month7
Month8
Month9
Month10
Month11
Month12
Month13
Month14
Month15
Month16
Month17
Month18
Month19
Month20
Month21
Month22
Month23
Month24
Month25
Month26
Month27
Month28
Month29
Month30
Month31
Month32
Month33
Month34
Month35
Month36
On-Demand
3yr Partial Upfront RI = 435 days to break even (40%)
3yr All Upfront RI = 409 days to breakeven (37%)
Breakeven Analysis for m3.medium (US-East, Linux) for NEW 3-year RIs !
Source: RightScale
After AWS RI model change of December 2, 2014
$- 
$100.00 
$200.00 
$300.00 
$400.00 
$500.00 
$600.00 
$700.00 
$800.00 
Month
1
Month
2
Month
3
Month
4
Month
5
Month
6
Month
7
Month
8
Month
9
Month
10
Month
11
Month
12
On-Demand
1yr Heavy RI = 226 days to breakeven (62%)
1yr Medium RI = 175 days to breakeven (48%)
1yr Light RI = Light COSTS MORE
Breakeven Analysis for m3.medium for OLD 1-year RIs !
Source: RightScale
After AWS RI model change of December 2, 2014
IMPACT OF PRICE DROPS
27
$0.58 
$0.50 
$0.45 
$0.28 
 $0.28 
Oct 2012
 Feb 2013
 Nov 2013
 Mar 2014
 Dec 2014
Cloud Price Drops
28
AWS On-Demand Price Drops
m3.xlarge US-East Linux
Drop of 51.7% in 26 months
Averages out to 23.9% in a year 
-10%!
-38%!
-13.8%!
No chg!
Launch!
Are we close to
another price
cut?
$- 
$100.00 
$200.00 
$300.00 
$400.00 
$500.00 
$600.00 
$700.00 
Month
1
Month
2
Month
3
Month
4
Month
5
Month
6
Month
7
Month
8
Month
9
Month
10
Month
11
Month
12
On-Demand- 261 days to breakeven
OD 24% cut in M7 - 286 days to breakeven
OD 24% cut in M1 - 346 days to breakeven
1yr No Upfront RI
Impact of Price Drops on Breakeven Analysis for m3.medium for 1-year RIs !
Source: RightScale
After AWS RI model change of December 2, 2014
29% savings
19% savings
6% savings
$- 
$200.00 
$400.00 
$600.00 
$800.00 
$1,000.00 
$1,200.00 
$1,400.00 
$1,600.00 
$1,800.00 
$2,000.00 
Month1
Month2
Month3
Month4
Month5
Month6
Month7
Month8
Month9
Month10
Month11
Month12
Month13
Month14
Month15
Month16
Month17
Month18
Month19
Month20
Month21
Month22
Month23
Month24
Month25
Month26
Month27
Month28
Month29
Month30
Month31
Month32
Month33
Month34
Month35
Month36
On-Demand
OD -24% yearly M7 - 515 days to breakeven
OD -24% yearly M1 - 638 days to breakeven
3yr Partial Upfront RI - 435 days to breakeven
Impact of Price Drops on Breakeven Analysis for m3.medium for 3-year RIs !
Source: RightScale
After AWS RI model change of December 2, 2014
60% savings
42% savings
33% savings
•  New instance types are introduced ocassionally
•  List some
•  What happens to prices and price/performance
•  What it means for RIs
What About New Instance Types?
31
•  T2 launched Jul 2014
•  R3 launched Apr 2014
•  I2 launched Dec 2013
•  C3 launched Nov 2013
•  M3 launched Oct 2012 
What about New Instance Types?
32
$0.58 
$0.50 
$0.45 
$0.28 
 $0.28 
0.68
0.64
0.52
 0.48
0.35
Oct 2009
 Sep 2012
 Oct 2012
 Feb 2013
 Nov 2013
 Mar 2014
 Dec 2014
Pricing on Instance Family Changes
33
AWS On-Demand Price Drops
m1.xlarge to m3.xlarge US-East Linux
Takeaways
•  New AWS RIs can save you money, but…
•  Future On-Demand price cuts can erode the differential
•  Do your breakeven analysis!
•  Try to consolidate instance families
•  Consider selling RIs you can’t use
•  Breakeven periods can be significant, so…
•  Work with your IT team to understand future usage 
•  Think about upcoming instance type changes
•  In the Google vs AWS price wars…
•  Google is getting more aggressive in cuts, backed by deep pockets
•  It’s unclear if/how AWS will respond over time
•  Longer commitments to AWS (3 yr RIs) can save you more
Q&A

Try Cloud Analytics 60-day Free Trial
www.rightscale.com/analytics


Read our blogs on AWS price changes and other cloud pricing topics
www.rightscale.com/blog/cloud-cost-analysis

More Related Content

PDF
Mastering the fundamentals of AWS billing 8-20-15
PDF
Proactive Cost Management for AWS Cloud
PDF
The Science of Saving with AWS Reserved Instances
PDF
The *New* Science Of Choosing AWS Reserved Instances
PDF
Strategies For Lasting Savings With AWS Reserved Instances
PDF
AWS Melbourne Cost Mgt. and Opti. Meetup - 20181109 - v2.2
PDF
AWS Cost Optimization in 5 Perspective
PDF
AWS Cost Optimization - JLM
Mastering the fundamentals of AWS billing 8-20-15
Proactive Cost Management for AWS Cloud
The Science of Saving with AWS Reserved Instances
The *New* Science Of Choosing AWS Reserved Instances
Strategies For Lasting Savings With AWS Reserved Instances
AWS Melbourne Cost Mgt. and Opti. Meetup - 20181109 - v2.2
AWS Cost Optimization in 5 Perspective
AWS Cost Optimization - JLM

What's hot (8)

PDF
AWS Cost Allocation best practices: How high-growth businesses succeed
PDF
ManageYourCostsAndGovernYourUsageOnAWS[1]
PDF
Optimizing EC2 usage on AWS
PDF
Cost Optimization in AWS
PPTX
Cost optimization - Don't overspend on AWS
PDF
Finding hidden waste in your AWS infrastructure - 2/11/16
PDF
AWS Cloud cost optimization
PDF
Optimizing AWS S3 storage costs and usage
AWS Cost Allocation best practices: How high-growth businesses succeed
ManageYourCostsAndGovernYourUsageOnAWS[1]
Optimizing EC2 usage on AWS
Cost Optimization in AWS
Cost optimization - Don't overspend on AWS
Finding hidden waste in your AWS infrastructure - 2/11/16
AWS Cloud cost optimization
Optimizing AWS S3 storage costs and usage
Ad

Similar to RightScale Webinar: Decoding AWS Reserved Instances (RIs) What It Means for Cloud Pricing (16)

PDF
How to Find and Fix Waste to Optimize Your Cloud Spend
PDF
12 Ways to Manage Cloud Costs and Optimize Cloud Spend
PDF
RightScale Webinar: Enterprise-Wide Cloud Cost Planning
PPTX
AWS Cost Optimization
PDF
RightScale Webinar - Tales From the Trenches: Understanding and Managing Clo...
PDF
AWS re:Invent re:Cap - 비용 최적화 - 모범사례와 아키텍처 설계 심화편 - 이원일
PDF
Best Practices for AWS Cloud Cost Optimization
PDF
How to Set Up a Cloud Cost Optimization Process for your Enterprise
PPTX
KGC 2013 AWS session
PDF
Bootcamp Recap: EC2 Reserved Instances
PPTX
Controle e otimização de custos na AWS
PDF
Google’s Committed Use Discounts vs. AWS Reserved Instances and More Ways to ...
PDF
AWS Reserved Instances: Turn your recommendations into purchases
PPTX
Webinar: AWS Partner Strategies For Success
PPTX
AWS Meet-up Atlanta: AWS Economics
PDF
COST OPTIMIZATION – SAVE 20-30% OF YOUR AWS MONTHLY COST – A REALITY OR FANTASY?
How to Find and Fix Waste to Optimize Your Cloud Spend
12 Ways to Manage Cloud Costs and Optimize Cloud Spend
RightScale Webinar: Enterprise-Wide Cloud Cost Planning
AWS Cost Optimization
RightScale Webinar - Tales From the Trenches: Understanding and Managing Clo...
AWS re:Invent re:Cap - 비용 최적화 - 모범사례와 아키텍처 설계 심화편 - 이원일
Best Practices for AWS Cloud Cost Optimization
How to Set Up a Cloud Cost Optimization Process for your Enterprise
KGC 2013 AWS session
Bootcamp Recap: EC2 Reserved Instances
Controle e otimização de custos na AWS
Google’s Committed Use Discounts vs. AWS Reserved Instances and More Ways to ...
AWS Reserved Instances: Turn your recommendations into purchases
Webinar: AWS Partner Strategies For Success
AWS Meet-up Atlanta: AWS Economics
COST OPTIMIZATION – SAVE 20-30% OF YOUR AWS MONTHLY COST – A REALITY OR FANTASY?
Ad

More from RightScale (20)

PDF
10 Must-Have Automated Cloud Policies for IT Governance
PDF
Kubernetes and Terraform in the Cloud: How RightScale Does DevOps
PDF
Optimize Software, SaaS, and Cloud with Flexera and RightScale
PDF
Prepare Your Enterprise Cloud Strategy for 2019: 7 Things to Think About Now
PDF
Multi-Cloud Management with RightScale CMP (Demo)
PDF
Comparing Cloud VM Types and Prices: AWS vs Azure vs Google vs IBM
PDF
How to Allocate and Report Cloud Costs with RightScale Optima
PDF
Should You Move Between AWS, Azure, or Google Clouds? Considerations, Pros an...
PDF
Using RightScale CMP with Cloud Provider Tools
PDF
Best Practices for Multi-Cloud Security and Compliance
PDF
Automating Multi-Cloud Policies for AWS, Azure, Google, and More
PDF
The 5 Stages of Cloud Management for Enterprises
PDF
9 Ways to Reduce Cloud Storage Costs
PDF
Serverless Comparison: AWS vs Azure vs Google vs IBM
PDF
Best Practices for Cloud Managed Services Providers: The Path to CMP Success
PDF
Cloud Storage Comparison: AWS vs Azure vs Google vs IBM
PDF
2018 Cloud Trends: RightScale State of the Cloud Report
PDF
Got a Multi-Cloud Strategy? How RightScale CMP Helps
PDF
How to Manage Cloud Costs with RightScale Optima
PDF
Top 10 Cloud Trends for 2018 and Actions You Can Take Now
10 Must-Have Automated Cloud Policies for IT Governance
Kubernetes and Terraform in the Cloud: How RightScale Does DevOps
Optimize Software, SaaS, and Cloud with Flexera and RightScale
Prepare Your Enterprise Cloud Strategy for 2019: 7 Things to Think About Now
Multi-Cloud Management with RightScale CMP (Demo)
Comparing Cloud VM Types and Prices: AWS vs Azure vs Google vs IBM
How to Allocate and Report Cloud Costs with RightScale Optima
Should You Move Between AWS, Azure, or Google Clouds? Considerations, Pros an...
Using RightScale CMP with Cloud Provider Tools
Best Practices for Multi-Cloud Security and Compliance
Automating Multi-Cloud Policies for AWS, Azure, Google, and More
The 5 Stages of Cloud Management for Enterprises
9 Ways to Reduce Cloud Storage Costs
Serverless Comparison: AWS vs Azure vs Google vs IBM
Best Practices for Cloud Managed Services Providers: The Path to CMP Success
Cloud Storage Comparison: AWS vs Azure vs Google vs IBM
2018 Cloud Trends: RightScale State of the Cloud Report
Got a Multi-Cloud Strategy? How RightScale CMP Helps
How to Manage Cloud Costs with RightScale Optima
Top 10 Cloud Trends for 2018 and Actions You Can Take Now

Recently uploaded (20)

PDF
Modernizing your data center with Dell and AMD
PDF
CIFDAQ's Market Insight: SEC Turns Pro Crypto
PDF
Advanced methodologies resolving dimensionality complications for autism neur...
PDF
The Rise and Fall of 3GPP – Time for a Sabbatical?
PDF
Bridging biosciences and deep learning for revolutionary discoveries: a compr...
PDF
Empathic Computing: Creating Shared Understanding
PDF
Build a system with the filesystem maintained by OSTree @ COSCUP 2025
PPTX
PA Analog/Digital System: The Backbone of Modern Surveillance and Communication
PDF
Advanced Soft Computing BINUS July 2025.pdf
PPT
“AI and Expert System Decision Support & Business Intelligence Systems”
PDF
Optimiser vos workloads AI/ML sur Amazon EC2 et AWS Graviton
PDF
KodekX | Application Modernization Development
PPTX
Detection-First SIEM: Rule Types, Dashboards, and Threat-Informed Strategy
PDF
GamePlan Trading System Review: Professional Trader's Honest Take
PDF
Spectral efficient network and resource selection model in 5G networks
PDF
Review of recent advances in non-invasive hemoglobin estimation
DOCX
The AUB Centre for AI in Media Proposal.docx
PDF
Electronic commerce courselecture one. Pdf
PDF
Reach Out and Touch Someone: Haptics and Empathic Computing
PDF
Shreyas Phanse Resume: Experienced Backend Engineer | Java • Spring Boot • Ka...
Modernizing your data center with Dell and AMD
CIFDAQ's Market Insight: SEC Turns Pro Crypto
Advanced methodologies resolving dimensionality complications for autism neur...
The Rise and Fall of 3GPP – Time for a Sabbatical?
Bridging biosciences and deep learning for revolutionary discoveries: a compr...
Empathic Computing: Creating Shared Understanding
Build a system with the filesystem maintained by OSTree @ COSCUP 2025
PA Analog/Digital System: The Backbone of Modern Surveillance and Communication
Advanced Soft Computing BINUS July 2025.pdf
“AI and Expert System Decision Support & Business Intelligence Systems”
Optimiser vos workloads AI/ML sur Amazon EC2 et AWS Graviton
KodekX | Application Modernization Development
Detection-First SIEM: Rule Types, Dashboards, and Threat-Informed Strategy
GamePlan Trading System Review: Professional Trader's Honest Take
Spectral efficient network and resource selection model in 5G networks
Review of recent advances in non-invasive hemoglobin estimation
The AUB Centre for AI in Media Proposal.docx
Electronic commerce courselecture one. Pdf
Reach Out and Touch Someone: Haptics and Empathic Computing
Shreyas Phanse Resume: Experienced Backend Engineer | Java • Spring Boot • Ka...

RightScale Webinar: Decoding AWS Reserved Instances (RIs) What It Means for Cloud Pricing

  • 2. •  Hassan Hosseini •  Product Manager, Cloud Analytics •  Kim Weins •  VP Marketing Speakers
  • 3. •  Understanding the New AWS Reserved Instances •  Comparing AWS to Google •  Breakeven Analysis •  Factoring in Future Price Drops Agenda
  • 4. What Happened •  Quick recap: What is a Reserved Instance •  A commitment from you to AWS to use a resource for a given period of time •  For that commitment, you get a discounted price (between 20% - 75%) •  You also get a capacity reservation •  It is not a VM. It is not a dedicated machine. •  What changed: •  Light, Medium and Heavy RIs were discontinued (you could see it coming) •  All Upfront, Partial Upfront and No Upfront RIs were introduced – discounts based on timing of payment •  All the new RIs are billed 24 hours a day – regardless of whether they are used •  What hasn’t changed: •  You still have to commit to an instance type, platform (OS), and Availability Zone •  You can request modifications to move AZs or change instance sizes within a family •  Terms are 1 and 3 years •  Your current RIs
  • 5. RI Savings Vary 5 RI price points and discounts are different for every region, instance type on OS!!! No Upfront Partial Upfront All Upfront Terms Available 1 year 1 year or 3 year 1 year or 3 year Savings for 1-Year RIs (Linux) 20-50% 25-58% 27-58% Savings for 3-Year RIs (Linux) N/A 41-74% 45-75% Hours Billed 24 hrs a day 24 hrs a day 24 hrs a day
  • 6. Effective Hourly Rate for RIs On! Demand! All! Upfront! $$$!$! Partial! Upfront! ! Medium! No! Upfront! Effective hourly rates of 1-year RIs Linux in US-East Assumes 100% utilization ________! Heavy! m3.medium! $.0700!$.0500! Light! $.0433!$.0425! $.0477! $.0766! 39% savings 38% savings 29% savings RI price points and discounts are different for every region, instance type on OS!!!
  • 7. •  You are now committing to the minimum number of instances used. •  You are charged hourly rate 24x7 regardless if you are using it •  The Effective Hourly Rate and the discounts published by AWS are based on 24x7 usage •  If you scale up and down, or turn off instances during weekends and holidays, you need to look closely at breakeven points. •  Example: There are 251 working days in 2014. If you turn off an instance on weekends or after hours, No Upfront is not worth it. •  Combine historic analysis with your knowledge of the future •  Can you commit to an instance type in a specific Availability Zone? •  Is you application’s architecture changing? (Dev/Ops team can help here) •  Instance type consolidation is more important •  Have a planned approach across your company on instance types and sizes •  Consolidation will make it easier to buy and use Ris to get highest discount and minimize waster •  Analysis on underutilized RIs (waste) becomes much more important Purchasing New AWS RIs 7
  • 8. •  Analyze past usage trends with Cloud Analytics What and How Many to Buy 8
  • 9. •  Forecasts future usage with Cloud Analytics What-if Scenarios 9
  • 10. •  How they are applied can be different to how you show/charge back •  Here is how they are applied: •  In Consolidated Billing, the purchasing account gets priority •  If you have bought the RI at the payer level, then: •  The oldest instance gets the reservation till it stops using it, then it goes to the next oldest •  The cheapest price is applied from all your reservations •  Here is how you can show/charge back •  Un-Blended rate: Show the costs as they are applied (instance 1 in account 1 used it for 3 hours, instance 1 in account 2 used it for 5 hours etc). Deal with the upfront cost too. •  Blended rate: Regardless of who used RIs or On-Demand instances, split sum based on usage. How RIs are Applied by AWS 10
  • 11. •  Volume Discounts •  $500K and above, you will get 5% discount •  $4M and above, you will get 10% discount •  These are within a region only (e.g. US-East). Discounts do not include Windows with SQL Server. •  Spot •  AWS selling their unused servers on an hourly bid price. •  If your application can handle an instance suddenly disappearing with no warning, this is an option. •  RI Marketplace •  Sell your unwanted Reserved Instances (US only). •  You set the upfront fee. This is what you will get if sold (minus 12% service charge). •  Cannot sell RIs that you have got volume discounts on. •  How long will it take to sell on the marketplace? •  Median of 14-19 days according to Jan 2014 analysis by http://blog.supplyframe.com •  RightScale customer – sold all RIs (primarily Heavy) within 2-4 weeks at residual value •  Buy shorter-term RIs Other Ways to Save on AWS Pricing 11
  • 12. AWS VS GOOGLE PRICING 12
  • 13. About Google Sustained-Use Discounts •  No commitment. The more you use an instance family during the month, the higher the discount. •  Hourly rate tiers down as you hit 25%, 50% and 75% of the time in a month. Usage Level % of Billing Cycle Incremental Rate % of On-Demand Baseline Sample Rate n1-standard-1 Total Cost 0-25% 100% $0.07 $13.02 25-50% 80% $0.056 $10.42 50-75% 60% $0.042 $7.81 75-100% 40% $0.028 $5.21 Monthly Cost at 100% usage $36.46
  • 14. Google SUD uses “Inferred Instances” 14 Source: Google!
  • 15. •  AWS No Upfront RIs vs. Google 100% SUD is closest to apples-to-apples •  Comparisons based on: •  US-East Region •  Linux operating system •  Assumes instances are running 24x7 •  No volume discount tiers •  Keep in mind that this is a price comparison only. The specific performance of each instance might vary. •  AWS high-memory and high-CPU instances have more memory along with an SSD, that balances Google’s lower price. Depending on workload, the high-CPU c3 series from AWS could provide a better price/performance option. AWS vs Google 15
  • 16. 16 Google Instance Type CPU Cores RAM AWS Instance Type CPU Cores RAM Google New On-Demand (per hour) AWS On-Demand (per hour) AWS Price vs. Google n1-standard-1 1 3.75 m3.medium 1 3.75 $ 0.063 $ 0.070 +11.11% n1-standard-2 2 7.5 m3.large 2 7.5 $ 0.126 $ 0.140 +11.11% n1-standard-4 4 15 m3.xlarge 4 15 $ 0.252 $ 0.280 +11.11% n1-standard-8 8 30 m3.2xlarge 8 30 $ 0.504 $ 0.560 +11.11% n1-highmem-2 2 13 r3.xlarge 2 15 $ 0.148 $ 0.175 +18.24% n1-highmem-4 4 26 r3.2xlarge 4 30.5 $ 0.296 $ 0.350 +18.24% n1-highmem-8 8 52 r3.4xlarge 8 61 $ 0.592 $ 0.700 +18.24% n1-highcpu-2 2 1.8 c3.large 2 3.75 $ 0.080 $ 0.105 +31.25% n1-highcpu-4 4 3.6 c3.xlarge 4 7.5 $ 0.160 $ 0.210 +31.25% n1-highcpu-8 8 7.2 c3.2xlarge 8 15 $ 0.320 $ 0.420 +31.25% n1-highcpu-16 16 14.4 c3.4xlarge 16 30 $ 0.640 $ 0.840 +31.25% Google vs. AWS On-Demand Pricing! Source: RightScale After GCE price cuts of October 1, 2014
  • 17. 17 Google Instance Type CPU Cores RAM AWS Instance Type CPU Cores RAM Google New 100% Usage ( per year) AWS 1-year No Upfront RI (per year) New AWS Price vs. Google n1-standard-1 1 3.75 m3.medium 1 3.75 $394 $438 +11.11% n1-standard-2 2 7.5 m3.large 2 7.5 $780 $876 +12.36% n1-standard-4 4 15 m3.xlarge 4 15 $1,551 $1,752 +12.99% n1-standard-8 8 30 m3.2xlarge 8 30 $3,094 $3,504 +13.25% n1-highmem-2 2 13 r3.xlarge 2 15 $911 $964 +5.77% n1-highmem-4 4 26 r3.2xlarge 4 30.5 $1,822 $1,927 +5.77% n1-highmem-8 8 52 r3.4xlarge 8 61 $3,635 $3,854 +6.02% n1-highcpu-2 2 1.8 c3.large 2 3.75 $491 $639 +30.36% n1-highcpu-4 4 3.6 c3.xlarge 4 7.5 $981 $1,279 +30.36% n1-highcpu-8 8 7.2 c3.2xlarge 8 15 $1,962 $2,558 +30.36% n1-highcpu-16 16 14.4 c3.4xlarge 16 30 $3,924 $5,116 +30.36% Google Sustained-Use vs. AWS 1-year No Upfront Reserved Instance Pricing! Source: RightScale After AWS RI model change of December 2, 2014
  • 18. 18 Google Instance Type CPU Cores RAM AWS Instance Type CPU Cores RAM Google New 100% Usage ( per year) AWS 1-year Partial Upfront RI (per year) New AWS Price vs. Google n1-standard-1 1 3.75 m3.medium 1 3.75 $394 $379 -3.78% n1-standard-2 2 7.5 m3.large 2 7.5 $780 $767 -1.57% n1-standard-4 4 15 m3.xlarge 4 15 $1,551 $1,534 -1.07% n1-standard-8 8 30 m3.2xlarge 8 30 $3,094 $3,051 -1.33% n1-highmem-2 2 13 r3.xlarge 2 15 $911 $830 -8.85% n1-highmem-4 4 26 r3.2xlarge 4 30.5 $1,822 $1,660 -8.89% n1-highmem-8 8 52 r3.4xlarge 8 61 $3,635 $3,320 -8.67% n1-highcpu-2 2 1.8 c3.large 2 3.75 $491 $554 +12.86% n1-highcpu-4 4 3.6 c3.xlarge 4 7.5 $981 $1,116 +13.75% n1-highcpu-8 8 7.2 c3.2xlarge 8 15 $1,962 $2,215 +12.90% n1-highcpu-16 16 14.4 c3.4xlarge 16 30 $3,924 $4,439 +13.10% Google Sustained-Use vs. AWS 1-year Partial Upfront Reserved Instance Pricing! Source: RightScale After AWS RI model change of December 2, 2014
  • 19. 19 Google Instance Type CPU Cores RAM AWS Instance Type CPU Cores RAM Google New 100% Usage ( per year) AWS 1-year Partial Upfront RI (per year) New AWS Price vs. Google n1-standard-1 1 3.75 m3.medium 1 3.75 $394 $372 -5.56% n1-standard-2 2 7.5 m3.large 2 7.5 $780 $751 -3.71% n1-standard-4 4 15 m3.xlarge 4 15 $1,551 $1,503 -3.05% n1-standard-8 8 30 m3.2xlarge 8 30 $3,094 $2.989 -3.34% n1-highmem-2 2 13 r3.xlarge 2 15 $911 $813 -10.77% n1-highmem-4 4 26 r3.2xlarge 4 30.5 $1,822 $1,626 -10.77% n1-highmem-8 8 52 r3.4xlarge 8 61 $3,635 $3,253 -10.53% n1-highcpu-2 2 1.8 c3.large 2 3.75 $491 $542 +10.54% n1-highcpu-4 4 3.6 c3.xlarge 4 7.5 $981 $1,093 +11.43% n1-highcpu-8 8 7.2 c3.2xlarge 8 15 $1,962 $2,170 +10.58% n1-highcpu-16 16 14.4 c3.4xlarge 16 30 $3,924 $4,350 +10.85% Google Sustained-Use vs. AWS 1-year All Upfront Reserved Instance Pricing! Source: RightScale After AWS RI model change of December 2, 2014
  • 20. 20 Google Instance Type CPU Cores RAM AWS Instance Type CPU Cores RAM Google New 100% Usage ( per year) AWS 3-year Partial Upfront RI (per year) New AWS Price vs. Google n1-standard-1 1 3.75 m3.medium 1 3.75 $394 $244 -38.22% n1-standard-2 2 7.5 m3.large 2 7.5 $780 $487 -37.52% n1-standard-4 4 15 m3.xlarge 4 15 $1,551 $974 -37.18% n1-standard-8 8 30 m3.2xlarge 8 30 $3,094 $1,948 -37.00% n1-highmem-2 2 13 r3.xlarge 2 15 $911 $572 -37.21% n1-highmem-4 4 26 r3.2xlarge 4 30.5 $1,822 $1,144 -37.21% n1-highmem-8 8 52 r3.4xlarge 8 61 $3,635 $2,288 -37.06% n1-highcpu-2 2 1.8 c3.large 2 3.75 $491 $362 -26.25% n1-highcpu-4 4 3.6 c3.xlarge 4 7.5 $981 $733 -25.27% n1-highcpu-8 8 7.2 c3.2xlarge 8 15 $1,962 $1,466 -25.31% n1-highcpu-16 16 14.4 c3.4xlarge 16 30 $3,924 $2,931 -25.31% Google Sustained-Use vs. AWS 3-year Partial Upfront Reserved Instance Pricing! Source: RightScale After AWS RI model change of December 2, 2014
  • 21. Things to Remember •  For AWS: •  High-CPU and high-mem have more memory and SSD •  You can get volume discounts (5% starting at $500K annual RI spend) •  You can choose Spot Pricing to save more •  You can use RI Marketplace to buy/sell RIs •  For Google: •  No commitment •  No lock-in to instance types or regions •  It’s automatic - no RI purchase decisions •  Per-minute pricing (minimum of 10 minutes) •  When on-demand price goes down, you get the benefit. •  They’ve lowered On-Demand prices 38% so far in 2014
  • 23. Calculating Breakeven for AWS RIs •  Breakeven Defined •  How long do I need to use an RI for it to be cheaper than On-demand? •  This is not a cash-flow analysis, but a cost analysis •  The Impact Upfront Commitment •  For all the new-style AWS Ris, you are committing to the full price of the RI upfront – the only difference is when you actually pay •  Unlike old Light and Medium RIs, with new AWS Ris you can’t stop paying the hourly fee even if you stop using the RI – no walkaway option
  • 24. $- $100.00 $200.00 $300.00 $400.00 $500.00 $600.00 $700.00 Month 1 Month 2 Month 3 Month 4 Month 5 Month 6 Month 7 Month 8 Month 9 Month 10 Month 11 Month 12 On-Demand 1yr No Upfront RI = 261 days to breakeven (72%) 1yr Partial Upfront RI = 226 days to breakeven (62%) 1yr All Upfront RI = 222 days to breakeven (61%) Breakeven Analysis for m3.medium (US-East, Linux) for NEW 1-year RIs ! Source: RightScale After AWS RI model change of December 2, 2014
  • 25. $- $200.00 $400.00 $600.00 $800.00 $1,000.00 $1,200.00 $1,400.00 $1,600.00 $1,800.00 $2,000.00 Month1 Month2 Month3 Month4 Month5 Month6 Month7 Month8 Month9 Month10 Month11 Month12 Month13 Month14 Month15 Month16 Month17 Month18 Month19 Month20 Month21 Month22 Month23 Month24 Month25 Month26 Month27 Month28 Month29 Month30 Month31 Month32 Month33 Month34 Month35 Month36 On-Demand 3yr Partial Upfront RI = 435 days to break even (40%) 3yr All Upfront RI = 409 days to breakeven (37%) Breakeven Analysis for m3.medium (US-East, Linux) for NEW 3-year RIs ! Source: RightScale After AWS RI model change of December 2, 2014
  • 26. $- $100.00 $200.00 $300.00 $400.00 $500.00 $600.00 $700.00 $800.00 Month 1 Month 2 Month 3 Month 4 Month 5 Month 6 Month 7 Month 8 Month 9 Month 10 Month 11 Month 12 On-Demand 1yr Heavy RI = 226 days to breakeven (62%) 1yr Medium RI = 175 days to breakeven (48%) 1yr Light RI = Light COSTS MORE Breakeven Analysis for m3.medium for OLD 1-year RIs ! Source: RightScale After AWS RI model change of December 2, 2014
  • 27. IMPACT OF PRICE DROPS 27
  • 28. $0.58 $0.50 $0.45 $0.28 $0.28 Oct 2012 Feb 2013 Nov 2013 Mar 2014 Dec 2014 Cloud Price Drops 28 AWS On-Demand Price Drops m3.xlarge US-East Linux Drop of 51.7% in 26 months Averages out to 23.9% in a year -10%! -38%! -13.8%! No chg! Launch! Are we close to another price cut?
  • 29. $- $100.00 $200.00 $300.00 $400.00 $500.00 $600.00 $700.00 Month 1 Month 2 Month 3 Month 4 Month 5 Month 6 Month 7 Month 8 Month 9 Month 10 Month 11 Month 12 On-Demand- 261 days to breakeven OD 24% cut in M7 - 286 days to breakeven OD 24% cut in M1 - 346 days to breakeven 1yr No Upfront RI Impact of Price Drops on Breakeven Analysis for m3.medium for 1-year RIs ! Source: RightScale After AWS RI model change of December 2, 2014 29% savings 19% savings 6% savings
  • 30. $- $200.00 $400.00 $600.00 $800.00 $1,000.00 $1,200.00 $1,400.00 $1,600.00 $1,800.00 $2,000.00 Month1 Month2 Month3 Month4 Month5 Month6 Month7 Month8 Month9 Month10 Month11 Month12 Month13 Month14 Month15 Month16 Month17 Month18 Month19 Month20 Month21 Month22 Month23 Month24 Month25 Month26 Month27 Month28 Month29 Month30 Month31 Month32 Month33 Month34 Month35 Month36 On-Demand OD -24% yearly M7 - 515 days to breakeven OD -24% yearly M1 - 638 days to breakeven 3yr Partial Upfront RI - 435 days to breakeven Impact of Price Drops on Breakeven Analysis for m3.medium for 3-year RIs ! Source: RightScale After AWS RI model change of December 2, 2014 60% savings 42% savings 33% savings
  • 31. •  New instance types are introduced ocassionally •  List some •  What happens to prices and price/performance •  What it means for RIs What About New Instance Types? 31
  • 32. •  T2 launched Jul 2014 •  R3 launched Apr 2014 •  I2 launched Dec 2013 •  C3 launched Nov 2013 •  M3 launched Oct 2012 What about New Instance Types? 32
  • 33. $0.58 $0.50 $0.45 $0.28 $0.28 0.68 0.64 0.52 0.48 0.35 Oct 2009 Sep 2012 Oct 2012 Feb 2013 Nov 2013 Mar 2014 Dec 2014 Pricing on Instance Family Changes 33 AWS On-Demand Price Drops m1.xlarge to m3.xlarge US-East Linux
  • 34. Takeaways •  New AWS RIs can save you money, but… •  Future On-Demand price cuts can erode the differential •  Do your breakeven analysis! •  Try to consolidate instance families •  Consider selling RIs you can’t use •  Breakeven periods can be significant, so… •  Work with your IT team to understand future usage •  Think about upcoming instance type changes •  In the Google vs AWS price wars… •  Google is getting more aggressive in cuts, backed by deep pockets •  It’s unclear if/how AWS will respond over time •  Longer commitments to AWS (3 yr RIs) can save you more
  • 35. Q&A Try Cloud Analytics 60-day Free Trial www.rightscale.com/analytics Read our blogs on AWS price changes and other cloud pricing topics www.rightscale.com/blog/cloud-cost-analysis