SlideShare uma empresa Scribd logo
1 de 25
Baixar para ler offline
A needle in the haystack: optimizing cloud
configurations for price-performance
by Stefano Doni
CTO @ akamas.io
The Problem
Cloud compute services offer overwhelming choices
EC2 instances cost ranges from $3.4 to $19482 per month (on demand)
https://www.slideshare.net/AmazonWebServices/deep-dive-on-amazon-ec2-instances-performance-optimization-best-practices-cmp307r1-
aws-reinvent-2018
Cloud storage services provide various price and
performance points
https://www.slideshare.net/AmazonWebServices/deep-dive-on-amazon-elastic-block-storage-amazon-ebs-stg310r1-aws-reinvent-2018
EBS cost ranges from $0.025 to $0.125 per GB-month + provisioned IOPS
Cloud compute instances and storage types are
interdependent
https://www.slideshare.net/AmazonWebServices/deep-dive-on-amazon-elastic-block-storage-amazon-ebs-stg310r1-aws-reinvent-2018
EC2 to EBS network can limit actual volume performance (e.g. IOPS)
bottleneck!
The current approaches
The model-based approach, aka cloud right sizing
recommendations
https://cloud.google.com/compute/docs/instances/apply-sizing-recommendations-for-instances
The experimental approach, aka load test your app
“There is no substitute for measuring the
performance of your entire application, because
application performance can be impacted by the
underlying infrastructure or by software and
architectural limitations.
We recommend application-level testing,
including the use of application profiling and load
testing tools and services”
https://aws.amazon.com/ec2/instance-types/
A bigger problem: same specs, different performance
across different cloud providers
“CockroachDB 2.1 achieves 40%
more throughput (tpmC) on
TPC-C when tested on AWS
using c5d.4xlarge than on GCP
via n1-standard-16.
We were shocked that AWS
offered such superior
performance”
Cockroach Labs
https://www.cockroachlabs.com/blog/2018_cloud_report/
Why current approaches can not assure optimal
application performance and low costs?
● May not consider end to end application performance
● May not capture hidden bottlenecks
● May not capture unique application / workload behaviour
● May not factor in cloud-specific platforms and
implementations (e.g. hypervisors, CPU architectures)
● Can’t scale to the sheer complexity of cloud options
The new AI-driven approach
Key capabilities
Powered by AI Automated Full-stack Goal-driven
A new vision: continuous and self-driving optimization
Configure Performance
Test
Measure
Goal
A real example: optimizing
MongoDB on AWS
The use case
Goal
Minimize price/performance of a MongoDB database hosted on AWS
Performance is throughput of the database (queries/sec), price is monthly
AWS price for the provisioned resources
Scenario
Akamas driving automated optimization including application load tests
Workflow to provision AWS EC2 and EBS resources as suggested by AI
engine
Optimization scope
AWS EC2 instances and EBS storage volumes powering MongoDB
Modeling the cloud cost-optimization problem
c5d.2xlarge
Instance family
Instance generation
Additional capabilities
Volume type
Instance size
Volume size
Volume IOPS
io1
70 GB
1000 IOPS
EC2
EBS
Results
AI-driven price-performance optimization results
Baseline configuration:
price/performance of
r4.large, gp2 70GB
Best configuration: -68%
price/performance
after 18 experiments
or approx 22 hours
Best configuration: for the same price, 3x throughput and -
90% latency
Price: - 2.9%
65.52 (best) vs 67.48 (baseline)
€/month
Throughput: +205%
7605 (best) vs 2493 (baseline)
query/sec
Latency (avg): -90%
1330 (best) vs 14575 (baseline)
milliseconds
How did AI achieve that? A look at the best configuration
Instance
Name
Use cases vCPUs Memory
(GiB)
Instance
Storage
Block
Storage (EBS)
r4.large
(baseline)
Memory
optimized
2 x Intel
Xeon E5-
2686
15.25 - gp2 70GB
m5d.large
(best)
General
purpose
2 x Custom
Intel Xeon
Platinum
8175M
8 1 x 150 GB
NVMe SSD
n/a
The best configuration for this workload is:
m5d.large
HW specs comparison
AI can find unusual configurations: AMD CPUs with half
memory can cut costs and still improve throughput
The cheapest configuration for this workload
is m5a.large
-24% cost with +12% throughput
Instance
Name
Use cases vCPUs Memory
(GiB)
Instance
Storage
Block Storage
(EBS)
r4.large
(baseline)
Memory
optimized
2 x Intel
Xeon
E5-2686
15.25 - gp2 70 GB
m5a.large
(cheapest)
Memory
optimized
2 x AMD
EPYC
8 - gp2 114 GB
HW specs comparison
Searching instances with EBS storage
Top 5 best configurations
r4.large
m5a.large
Memoryused
r4.large
m5a.large
ThroughputDebunking a common myth:
high resource usage != application performance bottleneck
… despite m5a.large (cheapest)
having half the memory of
r4.large (baseline)
Throughput +12% higher for the
m5a.large (cheapest) vs r4.large
(baseline) instance ...
Conclusions
Takeaways
● Technology landscape is becoming more and more complex
● Traditional approaches are not effective and can’t scale - significant
optimization opportunities are left on the table
● AI for IT optimization is required and can reach previously unthinkable
benefits, beyond what human experts can do
● In the cloud, 70% price/performance improvements are possible by
properly exploiting choices we have
● Cloud rightsizing recommendations may suggest higher price options
Q & A

Mais conteúdo relacionado

Mais procurados

AWS Presentation at JasperWorld APAC
AWS Presentation at JasperWorld APACAWS Presentation at JasperWorld APAC
AWS Presentation at JasperWorld APAC
Amazon Web Services
 
AWS Cloud Kata | Bangkok - Getting to Scale on AWS
AWS Cloud Kata | Bangkok - Getting to Scale on AWSAWS Cloud Kata | Bangkok - Getting to Scale on AWS
AWS Cloud Kata | Bangkok - Getting to Scale on AWS
Amazon Web Services
 
Optimizing Your Infrastructure Costs on AWS
Optimizing Your Infrastructure Costs on AWSOptimizing Your Infrastructure Costs on AWS
Optimizing Your Infrastructure Costs on AWS
Amazon Web Services
 
AWS Summit 2013 | Auckland - Big Data Analytics
AWS Summit 2013 | Auckland - Big Data AnalyticsAWS Summit 2013 | Auckland - Big Data Analytics
AWS Summit 2013 | Auckland - Big Data Analytics
Amazon Web Services
 

Mais procurados (20)

AWS Presentation at JasperWorld APAC
AWS Presentation at JasperWorld APACAWS Presentation at JasperWorld APAC
AWS Presentation at JasperWorld APAC
 
Predicting Costs on AWS
Predicting Costs on AWSPredicting Costs on AWS
Predicting Costs on AWS
 
Cloud cost optimization (AWS, GCP)
Cloud cost optimization (AWS, GCP)Cloud cost optimization (AWS, GCP)
Cloud cost optimization (AWS, GCP)
 
High Performance Computing with AWS
High Performance Computing with AWSHigh Performance Computing with AWS
High Performance Computing with AWS
 
AWS Cloud Kata | Bangkok - Getting to Scale on AWS
AWS Cloud Kata | Bangkok - Getting to Scale on AWSAWS Cloud Kata | Bangkok - Getting to Scale on AWS
AWS Cloud Kata | Bangkok - Getting to Scale on AWS
 
Optimizing Your AWS Applications and Usage to Reduce Costs
Optimizing Your AWS Applications and Usage to Reduce CostsOptimizing Your AWS Applications and Usage to Reduce Costs
Optimizing Your AWS Applications and Usage to Reduce Costs
 
Optimizing for Cost in the AWS Cloud - 5 Ways to Further Save - AWS Summit 20...
Optimizing for Cost in the AWS Cloud - 5 Ways to Further Save - AWS Summit 20...Optimizing for Cost in the AWS Cloud - 5 Ways to Further Save - AWS Summit 20...
Optimizing for Cost in the AWS Cloud - 5 Ways to Further Save - AWS Summit 20...
 
Optimizing Your Infrastructure Costs on AWS
Optimizing Your Infrastructure Costs on AWSOptimizing Your Infrastructure Costs on AWS
Optimizing Your Infrastructure Costs on AWS
 
Reducing Cost & Maximizing Efficiency: Tightening the Belt on AWS (CPN211) | ...
Reducing Cost & Maximizing Efficiency: Tightening the Belt on AWS (CPN211) | ...Reducing Cost & Maximizing Efficiency: Tightening the Belt on AWS (CPN211) | ...
Reducing Cost & Maximizing Efficiency: Tightening the Belt on AWS (CPN211) | ...
 
Spot at qubole
Spot at quboleSpot at qubole
Spot at qubole
 
Optimizing Costs and Efficiency of AWS Services
Optimizing Costs and Efficiency of AWS ServicesOptimizing Costs and Efficiency of AWS Services
Optimizing Costs and Efficiency of AWS Services
 
Optimising TCO with AWS at Websummit Dublin
Optimising TCO with AWS at Websummit DublinOptimising TCO with AWS at Websummit Dublin
Optimising TCO with AWS at Websummit Dublin
 
Aws cost strategies
Aws cost strategiesAws cost strategies
Aws cost strategies
 
AWS Vs Azure
AWS Vs AzureAWS Vs Azure
AWS Vs Azure
 
Cc
CcCc
Cc
 
EC2 Pricing Model (deck 0307 of the InfiniteSkills AWS course at http://bit.l...
EC2 Pricing Model (deck 0307 of the InfiniteSkills AWS course at http://bit.l...EC2 Pricing Model (deck 0307 of the InfiniteSkills AWS course at http://bit.l...
EC2 Pricing Model (deck 0307 of the InfiniteSkills AWS course at http://bit.l...
 
AWS Summit 2013 | Auckland - Big Data Analytics
AWS Summit 2013 | Auckland - Big Data AnalyticsAWS Summit 2013 | Auckland - Big Data Analytics
AWS Summit 2013 | Auckland - Big Data Analytics
 
(SOV203) Understanding AWS Storage Options | AWS re:Invent 2014
(SOV203) Understanding AWS Storage Options | AWS re:Invent 2014(SOV203) Understanding AWS Storage Options | AWS re:Invent 2014
(SOV203) Understanding AWS Storage Options | AWS re:Invent 2014
 
How to reduce hosting costs for Redis based applications on Java
How to reduce hosting costs for Redis based applications on JavaHow to reduce hosting costs for Redis based applications on Java
How to reduce hosting costs for Redis based applications on Java
 
Introduction to EC2 (AWS)
Introduction to EC2 (AWS)Introduction to EC2 (AWS)
Introduction to EC2 (AWS)
 

Semelhante a PAC 2019 virtual Stefano Doni

Improve your TCO and Optimise your Cloud Spend
Improve your TCO and Optimise your Cloud SpendImprove your TCO and Optimise your Cloud Spend
Improve your TCO and Optimise your Cloud Spend
Amazon Web Services
 

Semelhante a PAC 2019 virtual Stefano Doni (20)

IRJET- Cloud Cost Analyzer and Optimizer
IRJET- Cloud Cost Analyzer and OptimizerIRJET- Cloud Cost Analyzer and Optimizer
IRJET- Cloud Cost Analyzer and Optimizer
 
Deep Dive on Amazon EC2 Instances - January 2017 AWS Online Tech Talks
Deep Dive on Amazon EC2 Instances - January 2017 AWS Online Tech TalksDeep Dive on Amazon EC2 Instances - January 2017 AWS Online Tech Talks
Deep Dive on Amazon EC2 Instances - January 2017 AWS Online Tech Talks
 
AWS Summit Stockholm 2014 – B5 – The TCO of cloud applications
AWS Summit Stockholm 2014 – B5 – The TCO of cloud applicationsAWS Summit Stockholm 2014 – B5 – The TCO of cloud applications
AWS Summit Stockholm 2014 – B5 – The TCO of cloud applications
 
Deep Dive Amazon EC2
Deep Dive Amazon EC2Deep Dive Amazon EC2
Deep Dive Amazon EC2
 
Optimize Content Processing in the Cloud with GPU and Spot Instances
Optimize Content Processing in the Cloud with GPU and Spot InstancesOptimize Content Processing in the Cloud with GPU and Spot Instances
Optimize Content Processing in the Cloud with GPU and Spot Instances
 
Choosing the Right EC2 Instance and Applicable Use Cases - AWS June 2016 Webi...
Choosing the Right EC2 Instance and Applicable Use Cases - AWS June 2016 Webi...Choosing the Right EC2 Instance and Applicable Use Cases - AWS June 2016 Webi...
Choosing the Right EC2 Instance and Applicable Use Cases - AWS June 2016 Webi...
 
Deep Dive on Delivering Amazon EC2 Instance Performance
Deep Dive on Delivering Amazon EC2 Instance PerformanceDeep Dive on Delivering Amazon EC2 Instance Performance
Deep Dive on Delivering Amazon EC2 Instance Performance
 
Three Microsoft Azure SQL Managed Instances offered better SQL Server perform...
Three Microsoft Azure SQL Managed Instances offered better SQL Server perform...Three Microsoft Azure SQL Managed Instances offered better SQL Server perform...
Three Microsoft Azure SQL Managed Instances offered better SQL Server perform...
 
Deep Dive on Delivering Amazon EC2 Instance Performance
Deep Dive on Delivering Amazon EC2 Instance PerformanceDeep Dive on Delivering Amazon EC2 Instance Performance
Deep Dive on Delivering Amazon EC2 Instance Performance
 
StartPad Countdown 8 - Amazon Web Services and You
StartPad Countdown 8 - Amazon Web Services and YouStartPad Countdown 8 - Amazon Web Services and You
StartPad Countdown 8 - Amazon Web Services and You
 
Journey Through the AWS Cloud: Cost Optimisation
Journey Through the AWS Cloud: Cost OptimisationJourney Through the AWS Cloud: Cost Optimisation
Journey Through the AWS Cloud: Cost Optimisation
 
AWS APAC Webinar Series: How to Reduce Your Spend on AWS
AWS APAC Webinar Series: How to Reduce Your Spend on AWSAWS APAC Webinar Series: How to Reduce Your Spend on AWS
AWS APAC Webinar Series: How to Reduce Your Spend on AWS
 
How to Reduce your Spend on AWS
How to Reduce your Spend on AWSHow to Reduce your Spend on AWS
How to Reduce your Spend on AWS
 
B4 - The TCO of cloud applications
B4 - The TCO of cloud applicationsB4 - The TCO of cloud applications
B4 - The TCO of cloud applications
 
Improve your TCO and Optimise your Cloud Spend
Improve your TCO and Optimise your Cloud SpendImprove your TCO and Optimise your Cloud Spend
Improve your TCO and Optimise your Cloud Spend
 
AWS Summit London 2014 | Optimising TCO for the AWS Cloud (100)
AWS Summit London 2014 | Optimising TCO for the AWS Cloud (100)AWS Summit London 2014 | Optimising TCO for the AWS Cloud (100)
AWS Summit London 2014 | Optimising TCO for the AWS Cloud (100)
 
Re invent 2018 meetup presentation
Re invent 2018 meetup presentationRe invent 2018 meetup presentation
Re invent 2018 meetup presentation
 
Amazon EC2:Masterclass
Amazon EC2:MasterclassAmazon EC2:Masterclass
Amazon EC2:Masterclass
 
Get higher transaction throughput and better price/performance with an Amazon...
Get higher transaction throughput and better price/performance with an Amazon...Get higher transaction throughput and better price/performance with an Amazon...
Get higher transaction throughput and better price/performance with an Amazon...
 
4K Media Workflows on AWS
4K Media Workflows on AWS4K Media Workflows on AWS
4K Media Workflows on AWS
 

Mais de Neotys

Mais de Neotys (20)

PAC 2020 Santorin - Giovanni Paolo Gibilisco
PAC 2020 Santorin - Giovanni Paolo GibiliscoPAC 2020 Santorin - Giovanni Paolo Gibilisco
PAC 2020 Santorin - Giovanni Paolo Gibilisco
 
PAC 2020 Santorin - Stijn Schepers
PAC 2020 Santorin - Stijn SchepersPAC 2020 Santorin - Stijn Schepers
PAC 2020 Santorin - Stijn Schepers
 
PAC 2020 Santorin - Edoardo Varani
PAC 2020 Santorin - Edoardo VaraniPAC 2020 Santorin - Edoardo Varani
PAC 2020 Santorin - Edoardo Varani
 
PAC 2020 Santorin - Andreas Grabner
PAC 2020 Santorin - Andreas Grabner PAC 2020 Santorin - Andreas Grabner
PAC 2020 Santorin - Andreas Grabner
 
PAC 2020 Santorin - Vasilis Chatzinasios
PAC 2020 Santorin - Vasilis ChatzinasiosPAC 2020 Santorin - Vasilis Chatzinasios
PAC 2020 Santorin - Vasilis Chatzinasios
 
PAC 2020 Santorin - Gopalkrishnan Yadav
PAC 2020 Santorin - Gopalkrishnan YadavPAC 2020 Santorin - Gopalkrishnan Yadav
PAC 2020 Santorin - Gopalkrishnan Yadav
 
PAC 2020 Santorin - Hari Krishnan Ramachandran
PAC 2020 Santorin - Hari Krishnan RamachandranPAC 2020 Santorin - Hari Krishnan Ramachandran
PAC 2020 Santorin - Hari Krishnan Ramachandran
 
PAC 2020 Santorin - Joerek Van Gaalen
PAC 2020 Santorin - Joerek Van GaalenPAC 2020 Santorin - Joerek Van Gaalen
PAC 2020 Santorin - Joerek Van Gaalen
 
PAC 2020 Santorin - Ankur Jain
PAC 2020 Santorin - Ankur JainPAC 2020 Santorin - Ankur Jain
PAC 2020 Santorin - Ankur Jain
 
PAC 2020 Santorin - Stephen Townshend
PAC 2020 Santorin - Stephen TownshendPAC 2020 Santorin - Stephen Townshend
PAC 2020 Santorin - Stephen Townshend
 
PAC 2020 Santorin - Leandro Melendez
PAC 2020 Santorin - Leandro MelendezPAC 2020 Santorin - Leandro Melendez
PAC 2020 Santorin - Leandro Melendez
 
PAC 2019 virtual Stephen Townshend
PAC 2019 virtual Stephen TownshendPAC 2019 virtual Stephen Townshend
PAC 2019 virtual Stephen Townshend
 
PAC 2019 virtual Federico Toledo
PAC 2019 virtual Federico Toledo   PAC 2019 virtual Federico Toledo
PAC 2019 virtual Federico Toledo
 
PAC 2019 virtual Leandro Melendez
PAC 2019 virtual Leandro Melendez PAC 2019 virtual Leandro Melendez
PAC 2019 virtual Leandro Melendez
 
PAC 2019 virtual Mark Tomlinson
PAC 2019 virtual Mark TomlinsonPAC 2019 virtual Mark Tomlinson
PAC 2019 virtual Mark Tomlinson
 
PAC 2019 virtual Srivalli Aparna
PAC 2019 virtual Srivalli AparnaPAC 2019 virtual Srivalli Aparna
PAC 2019 virtual Srivalli Aparna
 
PAC 2019 virtual Reuben Rajan George
PAC 2019 virtual Reuben Rajan GeorgePAC 2019 virtual Reuben Rajan George
PAC 2019 virtual Reuben Rajan George
 
PAC 2019 virtual Joerek Van Gaalen
PAC 2019 virtual Joerek Van GaalenPAC 2019 virtual Joerek Van Gaalen
PAC 2019 virtual Joerek Van Gaalen
 
PAC 2019 virtual Hemalatha Murugesan
PAC 2019 virtual Hemalatha Murugesan  PAC 2019 virtual Hemalatha Murugesan
PAC 2019 virtual Hemalatha Murugesan
 
PAC 2019 virtual Bruno Audoux
PAC 2019 virtual Bruno Audoux PAC 2019 virtual Bruno Audoux
PAC 2019 virtual Bruno Audoux
 

Último

"Lesotho Leaps Forward: A Chronicle of Transformative Developments"
"Lesotho Leaps Forward: A Chronicle of Transformative Developments""Lesotho Leaps Forward: A Chronicle of Transformative Developments"
"Lesotho Leaps Forward: A Chronicle of Transformative Developments"
mphochane1998
 
1_Introduction + EAM Vocabulary + how to navigate in EAM.pdf
1_Introduction + EAM Vocabulary + how to navigate in EAM.pdf1_Introduction + EAM Vocabulary + how to navigate in EAM.pdf
1_Introduction + EAM Vocabulary + how to navigate in EAM.pdf
AldoGarca30
 
DeepFakes presentation : brief idea of DeepFakes
DeepFakes presentation : brief idea of DeepFakesDeepFakes presentation : brief idea of DeepFakes
DeepFakes presentation : brief idea of DeepFakes
MayuraD1
 
Cara Menggugurkan Sperma Yang Masuk Rahim Biyar Tidak Hamil
Cara Menggugurkan Sperma Yang Masuk Rahim Biyar Tidak HamilCara Menggugurkan Sperma Yang Masuk Rahim Biyar Tidak Hamil
Cara Menggugurkan Sperma Yang Masuk Rahim Biyar Tidak Hamil
Cara Menggugurkan Kandungan 087776558899
 
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
ssuser89054b
 

Último (20)

HOA1&2 - Module 3 - PREHISTORCI ARCHITECTURE OF KERALA.pptx
HOA1&2 - Module 3 - PREHISTORCI ARCHITECTURE OF KERALA.pptxHOA1&2 - Module 3 - PREHISTORCI ARCHITECTURE OF KERALA.pptx
HOA1&2 - Module 3 - PREHISTORCI ARCHITECTURE OF KERALA.pptx
 
COST-EFFETIVE and Energy Efficient BUILDINGS ptx
COST-EFFETIVE  and Energy Efficient BUILDINGS ptxCOST-EFFETIVE  and Energy Efficient BUILDINGS ptx
COST-EFFETIVE and Energy Efficient BUILDINGS ptx
 
Unleashing the Power of the SORA AI lastest leap
Unleashing the Power of the SORA AI lastest leapUnleashing the Power of the SORA AI lastest leap
Unleashing the Power of the SORA AI lastest leap
 
"Lesotho Leaps Forward: A Chronicle of Transformative Developments"
"Lesotho Leaps Forward: A Chronicle of Transformative Developments""Lesotho Leaps Forward: A Chronicle of Transformative Developments"
"Lesotho Leaps Forward: A Chronicle of Transformative Developments"
 
Thermal Engineering Unit - I & II . ppt
Thermal Engineering  Unit - I & II . pptThermal Engineering  Unit - I & II . ppt
Thermal Engineering Unit - I & II . ppt
 
Tamil Call Girls Bhayandar WhatsApp +91-9930687706, Best Service
Tamil Call Girls Bhayandar WhatsApp +91-9930687706, Best ServiceTamil Call Girls Bhayandar WhatsApp +91-9930687706, Best Service
Tamil Call Girls Bhayandar WhatsApp +91-9930687706, Best Service
 
1_Introduction + EAM Vocabulary + how to navigate in EAM.pdf
1_Introduction + EAM Vocabulary + how to navigate in EAM.pdf1_Introduction + EAM Vocabulary + how to navigate in EAM.pdf
1_Introduction + EAM Vocabulary + how to navigate in EAM.pdf
 
DeepFakes presentation : brief idea of DeepFakes
DeepFakes presentation : brief idea of DeepFakesDeepFakes presentation : brief idea of DeepFakes
DeepFakes presentation : brief idea of DeepFakes
 
DC MACHINE-Motoring and generation, Armature circuit equation
DC MACHINE-Motoring and generation, Armature circuit equationDC MACHINE-Motoring and generation, Armature circuit equation
DC MACHINE-Motoring and generation, Armature circuit equation
 
Unit 4_Part 1 CSE2001 Exception Handling and Function Template and Class Temp...
Unit 4_Part 1 CSE2001 Exception Handling and Function Template and Class Temp...Unit 4_Part 1 CSE2001 Exception Handling and Function Template and Class Temp...
Unit 4_Part 1 CSE2001 Exception Handling and Function Template and Class Temp...
 
PE 459 LECTURE 2- natural gas basic concepts and properties
PE 459 LECTURE 2- natural gas basic concepts and propertiesPE 459 LECTURE 2- natural gas basic concepts and properties
PE 459 LECTURE 2- natural gas basic concepts and properties
 
Bhubaneswar🌹Call Girls Bhubaneswar ❤Komal 9777949614 💟 Full Trusted CALL GIRL...
Bhubaneswar🌹Call Girls Bhubaneswar ❤Komal 9777949614 💟 Full Trusted CALL GIRL...Bhubaneswar🌹Call Girls Bhubaneswar ❤Komal 9777949614 💟 Full Trusted CALL GIRL...
Bhubaneswar🌹Call Girls Bhubaneswar ❤Komal 9777949614 💟 Full Trusted CALL GIRL...
 
Block diagram reduction techniques in control systems.ppt
Block diagram reduction techniques in control systems.pptBlock diagram reduction techniques in control systems.ppt
Block diagram reduction techniques in control systems.ppt
 
kiln thermal load.pptx kiln tgermal load
kiln thermal load.pptx kiln tgermal loadkiln thermal load.pptx kiln tgermal load
kiln thermal load.pptx kiln tgermal load
 
Computer Networks Basics of Network Devices
Computer Networks  Basics of Network DevicesComputer Networks  Basics of Network Devices
Computer Networks Basics of Network Devices
 
Cara Menggugurkan Sperma Yang Masuk Rahim Biyar Tidak Hamil
Cara Menggugurkan Sperma Yang Masuk Rahim Biyar Tidak HamilCara Menggugurkan Sperma Yang Masuk Rahim Biyar Tidak Hamil
Cara Menggugurkan Sperma Yang Masuk Rahim Biyar Tidak Hamil
 
Introduction to Serverless with AWS Lambda
Introduction to Serverless with AWS LambdaIntroduction to Serverless with AWS Lambda
Introduction to Serverless with AWS Lambda
 
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
 
AIRCANVAS[1].pdf mini project for btech students
AIRCANVAS[1].pdf mini project for btech studentsAIRCANVAS[1].pdf mini project for btech students
AIRCANVAS[1].pdf mini project for btech students
 
GEAR TRAIN- BASIC CONCEPTS AND WORKING PRINCIPLE
GEAR TRAIN- BASIC CONCEPTS AND WORKING PRINCIPLEGEAR TRAIN- BASIC CONCEPTS AND WORKING PRINCIPLE
GEAR TRAIN- BASIC CONCEPTS AND WORKING PRINCIPLE
 

PAC 2019 virtual Stefano Doni

  • 1. A needle in the haystack: optimizing cloud configurations for price-performance by Stefano Doni CTO @ akamas.io
  • 3. Cloud compute services offer overwhelming choices EC2 instances cost ranges from $3.4 to $19482 per month (on demand) https://www.slideshare.net/AmazonWebServices/deep-dive-on-amazon-ec2-instances-performance-optimization-best-practices-cmp307r1- aws-reinvent-2018
  • 4. Cloud storage services provide various price and performance points https://www.slideshare.net/AmazonWebServices/deep-dive-on-amazon-elastic-block-storage-amazon-ebs-stg310r1-aws-reinvent-2018 EBS cost ranges from $0.025 to $0.125 per GB-month + provisioned IOPS
  • 5. Cloud compute instances and storage types are interdependent https://www.slideshare.net/AmazonWebServices/deep-dive-on-amazon-elastic-block-storage-amazon-ebs-stg310r1-aws-reinvent-2018 EC2 to EBS network can limit actual volume performance (e.g. IOPS) bottleneck!
  • 7. The model-based approach, aka cloud right sizing recommendations https://cloud.google.com/compute/docs/instances/apply-sizing-recommendations-for-instances
  • 8. The experimental approach, aka load test your app “There is no substitute for measuring the performance of your entire application, because application performance can be impacted by the underlying infrastructure or by software and architectural limitations. We recommend application-level testing, including the use of application profiling and load testing tools and services” https://aws.amazon.com/ec2/instance-types/
  • 9. A bigger problem: same specs, different performance across different cloud providers “CockroachDB 2.1 achieves 40% more throughput (tpmC) on TPC-C when tested on AWS using c5d.4xlarge than on GCP via n1-standard-16. We were shocked that AWS offered such superior performance” Cockroach Labs https://www.cockroachlabs.com/blog/2018_cloud_report/
  • 10. Why current approaches can not assure optimal application performance and low costs? ● May not consider end to end application performance ● May not capture hidden bottlenecks ● May not capture unique application / workload behaviour ● May not factor in cloud-specific platforms and implementations (e.g. hypervisors, CPU architectures) ● Can’t scale to the sheer complexity of cloud options
  • 11. The new AI-driven approach
  • 12. Key capabilities Powered by AI Automated Full-stack Goal-driven
  • 13. A new vision: continuous and self-driving optimization Configure Performance Test Measure Goal
  • 14. A real example: optimizing MongoDB on AWS
  • 15. The use case Goal Minimize price/performance of a MongoDB database hosted on AWS Performance is throughput of the database (queries/sec), price is monthly AWS price for the provisioned resources Scenario Akamas driving automated optimization including application load tests Workflow to provision AWS EC2 and EBS resources as suggested by AI engine Optimization scope AWS EC2 instances and EBS storage volumes powering MongoDB
  • 16. Modeling the cloud cost-optimization problem c5d.2xlarge Instance family Instance generation Additional capabilities Volume type Instance size Volume size Volume IOPS io1 70 GB 1000 IOPS EC2 EBS
  • 18. AI-driven price-performance optimization results Baseline configuration: price/performance of r4.large, gp2 70GB Best configuration: -68% price/performance after 18 experiments or approx 22 hours
  • 19. Best configuration: for the same price, 3x throughput and - 90% latency Price: - 2.9% 65.52 (best) vs 67.48 (baseline) €/month Throughput: +205% 7605 (best) vs 2493 (baseline) query/sec Latency (avg): -90% 1330 (best) vs 14575 (baseline) milliseconds
  • 20. How did AI achieve that? A look at the best configuration Instance Name Use cases vCPUs Memory (GiB) Instance Storage Block Storage (EBS) r4.large (baseline) Memory optimized 2 x Intel Xeon E5- 2686 15.25 - gp2 70GB m5d.large (best) General purpose 2 x Custom Intel Xeon Platinum 8175M 8 1 x 150 GB NVMe SSD n/a The best configuration for this workload is: m5d.large HW specs comparison
  • 21. AI can find unusual configurations: AMD CPUs with half memory can cut costs and still improve throughput The cheapest configuration for this workload is m5a.large -24% cost with +12% throughput Instance Name Use cases vCPUs Memory (GiB) Instance Storage Block Storage (EBS) r4.large (baseline) Memory optimized 2 x Intel Xeon E5-2686 15.25 - gp2 70 GB m5a.large (cheapest) Memory optimized 2 x AMD EPYC 8 - gp2 114 GB HW specs comparison Searching instances with EBS storage Top 5 best configurations
  • 22. r4.large m5a.large Memoryused r4.large m5a.large ThroughputDebunking a common myth: high resource usage != application performance bottleneck … despite m5a.large (cheapest) having half the memory of r4.large (baseline) Throughput +12% higher for the m5a.large (cheapest) vs r4.large (baseline) instance ...
  • 24. Takeaways ● Technology landscape is becoming more and more complex ● Traditional approaches are not effective and can’t scale - significant optimization opportunities are left on the table ● AI for IT optimization is required and can reach previously unthinkable benefits, beyond what human experts can do ● In the cloud, 70% price/performance improvements are possible by properly exploiting choices we have ● Cloud rightsizing recommendations may suggest higher price options
  • 25. Q & A