SlideShare a Scribd company logo
Power Comparison
of Cloud Data Center Architectures
Pietro Ruiu
Andrea Bianco, Paolo Giaccone
Claudio Fiandrino, Dzmitry Kliazovich
13th Italian Networking Workshop: San Candido, Italy
January 13 - 15, 2016
The quest for green data centers
• Data centers are the new polluters of 21st
century
– in 2012, accounted for 15% of the global ICT
energy consumption
– expected to increase in the next years
• Data center consumption
– 75% ICT equipment
• powering and cooling
• mostly due to servers
– 25% power distribution and facility operations
• Strong interest in designing and operating
data centers with higher energy efficiency
– not only to reduce OPEX costs
2
Key question
• Given
• a data center topology
• a power consumption profile for each ICT device
• Define the min-power job allocation policy
• Evaluate power consumption as function of the data center
load
Classical questions
• Given
• a power consumption profile for each ICT device
• a generic job allocation policy
• Compare the power consumption behavior in function of the
data center topology
Our question
3
Data center model
4
Data center network (DCN)
Power
Load
Power
Load
Power
Load
Servers
ICT device Power profile
Local vs global energy proportionality
• consume proportional to the load
• consume (and pay) only if really needed
Ideal energy proportionality
• Constant power (CONST)
• Full Energy Proportional (FEP)
• Linear (LIN)
Local power consumption for single device
• maybe very different from local power consumption
Global power consumption for overall system
5
CONST
Power
Load
FEP
Power
Load
LIN
Power
Load
Power
Load
Global power consumption
6
• N resources/devices to be allocated for a set of requests/VMs
• power consumption profile for each resource/device
• allocation policy
• consolidate: activate the minimum number of resources
• load-balance: distribute load across the resources
Resource allocation
• depends on granularity of the resources (i.e. the value of N)
• depends on allocation policy
Overall power consumption
Global Power Consumption
7
Local power
consumption
Consolidate policy Load-balance policy
Normalized power = Power / Load
FEP
Power
Load
Power Load Load
Norm.Power
Power
Load Load
Norm.Power
CONST
Power
Load
N
Load
Power
N
Load
Power
N Load
Norm.Power N
Load
Norm.Power
• Local energy proportionality implies global energy proportionality
• If N is enough large, consolidate policy reaches global energy
proportionality for CONST local power
Local and Global Energy Proportionality
Our contributions
• Network-aware min-power VM online
allocation policy
• Flow-level C++ simulator of data centers
• Power comparison of different network
topologies
8
• Energy profiles for each
ICT device (switch, link,
server)
• DCN topology
• VM arrival process
Flow-level
simulator
• Global power
consumption
• Load on each ICT
device
• VM blocking
probabilityVM allocation
policy
DCN topologies
• traditional Clos-based switch topologies
– classical 2, 3 tiers
– Jupiter
• Google’s disclosed DCN architecture
• “Jupiter Rising: A Decade of Clos Topologies and Centralized Control in
Google’s Datacenter Network”, ACM SIGCOMM CCR, Oct. 2015
9
• Switches: 10 core sw – 18 TOR sw
• Servers 180
• Total ICT devices: 208 nodes
CORE
TOR
10Gbps40Gbps
2-tiers DCN
10
• 3 core sw – 6 aggregation sw – 18 TOR sw
• 180 servers – 27 switches – 207 nodes
CORE
AGGREGATION
TOR
10Gbps40Gbps40Gbps
3-tiers DCN
11
• 24 spine sw – 16 aggregation sw – 16 TOR sw
• 192 servers – 44 switches – 236 nodes
= 4p@40Gbps (or 16p@10Gbps)
SPINEAGGREGATIONTOR
10Gbps10Gbps40Gbps40Gbps40Gbps
MB MB MB MB
Jupiter-like DCN
12
Online VM allocation policy
13
• for each VM, select a server at random
• connect through the minimum incremental DCN power
• load-balance on the servers
RSS (Random Server Selection)
• for each VM, select the server with minimum
incremental power (server + DCN)
• consolidate VMs in the same server, in the same rack, in
closeby racks, etc
• variant of min-cost Dijkstra algorithm
MNP (Minimum Network Power)
VM generation
• time is slotted
• at each timelot, a new VM arrives and must
communicate B bps to a previously randomly allocated
VM
– B is randomly chosen
– destination VM is chosen with Bernoulli trials
• simulation can run until saturating the data center
14
VM1 VM2 VM3 VM4 VM5
RSS (Random Server Selection)
• small datacenter (180-192 servers)
• Jupiter appears to be the most energy proportional
– due to the larger number of switches (44 vs 27-28)
15
0
5
10
15
20
25
20 30 40 50 60 70 80 90 100
NormalizedpowerperVM[W]
Data Center load [%]
3-Tier
LIN
FEP
CONST
0
5
10
15
20
25
20 30 40 50 60 70 80 90 100
Data Center load [%]
2-Tier
LIN
FEP
CONST
0
5
10
15
20
25
20 30 40 50 60 70 80 90 100
Data Center load [%]
Jupiter
LIN
FEP
CONST
MNP (Minimum Network Power)
16
• small datacenter (180-192 servers)
• MNP allows to achieve global energy-proportionality
• under FEP, power jumps due to abrupt activation of new layers
0
0.5
1
1.5
2
2.5
3
20 30 40 50 60 70 80 90100
NormalizedpowerperVM[W]
Data Center load [%]
3-Tier
LIN
FEP
CONST
0
0.5
1
1.5
2
2.5
3
20 30 40 50 60 70 80 90 100
Data Center load [%]
2-Tier
LIN
FEP
CONST
0
0.5
1
1.5
2
2.5
3
20 30 40 50 60 70 80 90 100
Data Center load [%]
Jupiter
LIN
FEP
CONST
Conclusions
• Global energy proportionality of an overall data center
depends on
• local power profile of each device
• topology (number of devices)
• VM allocation policy
Take-home message
• consider large topologies with 10,000 servers
• compare data center networks given the same bisection
bandwidth
• consider the allocation of clusters of VMs
Future works
17

More Related Content

What's hot

Job sequence scheduling for cloud computing
Job sequence scheduling for cloud computingJob sequence scheduling for cloud computing
Job sequence scheduling for cloud computing
Samruddhi Gaikwad
 
Cloud computing(bit mesra kolkata extn.)
Cloud computing(bit mesra kolkata extn.)Cloud computing(bit mesra kolkata extn.)
Cloud computing(bit mesra kolkata extn.)ASHUTOSH KUMAR
 
Energy efficient VM placement - OpenStack Summit Vancouver May 2015
Energy efficient VM placement - OpenStack Summit Vancouver May 2015Energy efficient VM placement - OpenStack Summit Vancouver May 2015
Energy efficient VM placement - OpenStack Summit Vancouver May 2015
Kurt Garloff
 
Energy and carbon efficient placement of virtual machines in distributed clou...
Energy and carbon efficient placement of virtual machines in distributed clou...Energy and carbon efficient placement of virtual machines in distributed clou...
Energy and carbon efficient placement of virtual machines in distributed clou...
Pradeeban Kathiravelu, Ph.D.
 
Base paper ppt-. A load balancing model based on cloud partitioning for the ...
Base paper ppt-. A  load balancing model based on cloud partitioning for the ...Base paper ppt-. A  load balancing model based on cloud partitioning for the ...
Base paper ppt-. A load balancing model based on cloud partitioning for the ...
Lavanya Vigrahala
 
A Comparative Study between Honeybee Foraging Behaviour Algorithm and Round ...
A Comparative Study between Honeybee Foraging Behaviour Algorithm and  Round ...A Comparative Study between Honeybee Foraging Behaviour Algorithm and  Round ...
A Comparative Study between Honeybee Foraging Behaviour Algorithm and Round ...
sondhicse
 
LOAD BALANCING ALGORITHM TO IMPROVE RESPONSE TIME ON CLOUD COMPUTING
LOAD BALANCING ALGORITHM TO IMPROVE RESPONSE TIME ON CLOUD COMPUTINGLOAD BALANCING ALGORITHM TO IMPROVE RESPONSE TIME ON CLOUD COMPUTING
LOAD BALANCING ALGORITHM TO IMPROVE RESPONSE TIME ON CLOUD COMPUTING
ijccsa
 
Vm consolidation for energy efficient cloud computing
Vm consolidation for energy efficient cloud computingVm consolidation for energy efficient cloud computing
Vm consolidation for energy efficient cloud computing
Hemanandhini Ganesan
 
Run-time power management in cloud and containerized environments
Run-time power management in cloud and containerized environmentsRun-time power management in cloud and containerized environments
Run-time power management in cloud and containerized environments
NECST Lab @ Politecnico di Milano
 
A presentation on economic load dispatch
A presentation on economic load dispatchA presentation on economic load dispatch
A presentation on economic load dispatch
souravsahoo28
 
09 mikoski pv-mismatch_pvpmc-8_20170509_r5
09 mikoski pv-mismatch_pvpmc-8_20170509_r509 mikoski pv-mismatch_pvpmc-8_20170509_r5
09 mikoski pv-mismatch_pvpmc-8_20170509_r5
Sandia National Laboratories: Energy & Climate: Renewables
 
A Study on Task Scheduling in Could Data Centers for Energy Efficacy
A Study on Task Scheduling in Could Data Centers for Energy Efficacy A Study on Task Scheduling in Could Data Centers for Energy Efficacy
A Study on Task Scheduling in Could Data Centers for Energy Efficacy
Ehsan Sharifi
 
Cluster computing
Cluster computingCluster computing
Cluster computingbrainbix
 
Simulation of Heterogeneous Cloud Infrastructures
Simulation of Heterogeneous Cloud InfrastructuresSimulation of Heterogeneous Cloud Infrastructures
Simulation of Heterogeneous Cloud Infrastructures
CloudLightning
 
Elastic Tree: Saving Energy in Data Center Networks
Elastic Tree: Saving Energy in Data Center NetworksElastic Tree: Saving Energy in Data Center Networks
Elastic Tree: Saving Energy in Data Center NetworksAbhishek Sutrave
 
Joint Approach of Routing , Rate Adaptation and Power Control in Wireless Mes...
Joint Approach of Routing , Rate Adaptation and Power Control in Wireless Mes...Joint Approach of Routing , Rate Adaptation and Power Control in Wireless Mes...
Joint Approach of Routing , Rate Adaptation and Power Control in Wireless Mes...
Manjunath Badiger
 
DQN Variants: A quick glance
DQN Variants: A quick glanceDQN Variants: A quick glance
DQN Variants: A quick glance
Tejas Kotha
 
Joint Approach of Routing , Rate Adaptation and Power Control in Wireless Mes...
Joint Approach of Routing , Rate Adaptation and Power Control in Wireless Mes...Joint Approach of Routing , Rate Adaptation and Power Control in Wireless Mes...
Joint Approach of Routing , Rate Adaptation and Power Control in Wireless Mes...
Manjunath Badiger
 
Survey: An Optimized Energy Consumption of Resources in Cloud Data Centers
Survey: An Optimized Energy Consumption of Resources in Cloud Data CentersSurvey: An Optimized Energy Consumption of Resources in Cloud Data Centers
Survey: An Optimized Energy Consumption of Resources in Cloud Data Centers
IJCSIS Research Publications
 

What's hot (20)

Job sequence scheduling for cloud computing
Job sequence scheduling for cloud computingJob sequence scheduling for cloud computing
Job sequence scheduling for cloud computing
 
Cloud computing(bit mesra kolkata extn.)
Cloud computing(bit mesra kolkata extn.)Cloud computing(bit mesra kolkata extn.)
Cloud computing(bit mesra kolkata extn.)
 
Energy efficient VM placement - OpenStack Summit Vancouver May 2015
Energy efficient VM placement - OpenStack Summit Vancouver May 2015Energy efficient VM placement - OpenStack Summit Vancouver May 2015
Energy efficient VM placement - OpenStack Summit Vancouver May 2015
 
Energy and carbon efficient placement of virtual machines in distributed clou...
Energy and carbon efficient placement of virtual machines in distributed clou...Energy and carbon efficient placement of virtual machines in distributed clou...
Energy and carbon efficient placement of virtual machines in distributed clou...
 
Base paper ppt-. A load balancing model based on cloud partitioning for the ...
Base paper ppt-. A  load balancing model based on cloud partitioning for the ...Base paper ppt-. A  load balancing model based on cloud partitioning for the ...
Base paper ppt-. A load balancing model based on cloud partitioning for the ...
 
A Comparative Study between Honeybee Foraging Behaviour Algorithm and Round ...
A Comparative Study between Honeybee Foraging Behaviour Algorithm and  Round ...A Comparative Study between Honeybee Foraging Behaviour Algorithm and  Round ...
A Comparative Study between Honeybee Foraging Behaviour Algorithm and Round ...
 
LOAD BALANCING ALGORITHM TO IMPROVE RESPONSE TIME ON CLOUD COMPUTING
LOAD BALANCING ALGORITHM TO IMPROVE RESPONSE TIME ON CLOUD COMPUTINGLOAD BALANCING ALGORITHM TO IMPROVE RESPONSE TIME ON CLOUD COMPUTING
LOAD BALANCING ALGORITHM TO IMPROVE RESPONSE TIME ON CLOUD COMPUTING
 
Vm consolidation for energy efficient cloud computing
Vm consolidation for energy efficient cloud computingVm consolidation for energy efficient cloud computing
Vm consolidation for energy efficient cloud computing
 
Run-time power management in cloud and containerized environments
Run-time power management in cloud and containerized environmentsRun-time power management in cloud and containerized environments
Run-time power management in cloud and containerized environments
 
A presentation on economic load dispatch
A presentation on economic load dispatchA presentation on economic load dispatch
A presentation on economic load dispatch
 
09 mikoski pv-mismatch_pvpmc-8_20170509_r5
09 mikoski pv-mismatch_pvpmc-8_20170509_r509 mikoski pv-mismatch_pvpmc-8_20170509_r5
09 mikoski pv-mismatch_pvpmc-8_20170509_r5
 
A Study on Task Scheduling in Could Data Centers for Energy Efficacy
A Study on Task Scheduling in Could Data Centers for Energy Efficacy A Study on Task Scheduling in Could Data Centers for Energy Efficacy
A Study on Task Scheduling in Could Data Centers for Energy Efficacy
 
Cluster computing
Cluster computingCluster computing
Cluster computing
 
Simulation of Heterogeneous Cloud Infrastructures
Simulation of Heterogeneous Cloud InfrastructuresSimulation of Heterogeneous Cloud Infrastructures
Simulation of Heterogeneous Cloud Infrastructures
 
Elastic Tree: Saving Energy in Data Center Networks
Elastic Tree: Saving Energy in Data Center NetworksElastic Tree: Saving Energy in Data Center Networks
Elastic Tree: Saving Energy in Data Center Networks
 
Project_Poster
Project_PosterProject_Poster
Project_Poster
 
Joint Approach of Routing , Rate Adaptation and Power Control in Wireless Mes...
Joint Approach of Routing , Rate Adaptation and Power Control in Wireless Mes...Joint Approach of Routing , Rate Adaptation and Power Control in Wireless Mes...
Joint Approach of Routing , Rate Adaptation and Power Control in Wireless Mes...
 
DQN Variants: A quick glance
DQN Variants: A quick glanceDQN Variants: A quick glance
DQN Variants: A quick glance
 
Joint Approach of Routing , Rate Adaptation and Power Control in Wireless Mes...
Joint Approach of Routing , Rate Adaptation and Power Control in Wireless Mes...Joint Approach of Routing , Rate Adaptation and Power Control in Wireless Mes...
Joint Approach of Routing , Rate Adaptation and Power Control in Wireless Mes...
 
Survey: An Optimized Energy Consumption of Resources in Cloud Data Centers
Survey: An Optimized Energy Consumption of Resources in Cloud Data CentersSurvey: An Optimized Energy Consumption of Resources in Cloud Data Centers
Survey: An Optimized Energy Consumption of Resources in Cloud Data Centers
 

Viewers also liked

Travelling to the far side of Andromeda
Travelling to the far side of AndromedaTravelling to the far side of Andromeda
Travelling to the far side of Andromeda
Jose Miguel Esparza
 
Modeling and Optimization of Resource Allocation in Cloud [PhD Thesis Progres...
Modeling and Optimization of Resource Allocation in Cloud [PhD Thesis Progres...Modeling and Optimization of Resource Allocation in Cloud [PhD Thesis Progres...
Modeling and Optimization of Resource Allocation in Cloud [PhD Thesis Progres...
AtakanAral
 
Green cloud computing
Green cloud computingGreen cloud computing
Green cloud computing
Karishma Patro
 
Plan your Log Platform at Google Cloud Platform
Plan your Log Platform at Google Cloud PlatformPlan your Log Platform at Google Cloud Platform
Plan your Log Platform at Google Cloud Platform
Simon Su
 
A Look Inside Google’s Data Center Networks
A Look Inside Google’s Data Center NetworksA Look Inside Google’s Data Center Networks
A Look Inside Google’s Data Center Networks
Ryousei Takano
 
AI for business: Capire l'opportunità
AI for business: Capire l'opportunitàAI for business: Capire l'opportunità
AI for business: Capire l'opportunità
MeetupDataScienceRoma
 
Outsourcing Services: What is right for your organisation
Outsourcing Services: What is right for your organisationOutsourcing Services: What is right for your organisation
Outsourcing Services: What is right for your organisation
Vincent Kwon
 
ONS content extraction
ONS content extractionONS content extraction
ONS content extractionKellyCheah
 
presentationGAATT
presentationGAATTpresentationGAATT
presentationGAATTKellyCheah
 
Serverless Data Architecture at scale on Google Cloud Platform
Serverless Data Architecture at scale on Google Cloud PlatformServerless Data Architecture at scale on Google Cloud Platform
Serverless Data Architecture at scale on Google Cloud Platform
MeetupDataScienceRoma
 
Introduction to Google Cloud Platform Technologies
Introduction to Google Cloud Platform TechnologiesIntroduction to Google Cloud Platform Technologies
Introduction to Google Cloud Platform Technologies
Chris Schalk
 
Green cloud computing
Green cloud computingGreen cloud computing
Green cloud computing
Ram kumar
 
Green cloud computing
Green cloud computingGreen cloud computing
Green cloud computing
Shreyas Khare
 
Google Cloud Platform Empowers TensorFlow and Machine Learning
Google Cloud Platform Empowers TensorFlow and Machine LearningGoogle Cloud Platform Empowers TensorFlow and Machine Learning
Google Cloud Platform Empowers TensorFlow and Machine Learning
DataWorks Summit/Hadoop Summit
 
Google Cloud Platform
Google Cloud Platform Google Cloud Platform
Google Cloud Platform
Francesco Marchitelli
 
Understanding cloud with Google Cloud Platform
Understanding cloud with Google Cloud PlatformUnderstanding cloud with Google Cloud Platform
Understanding cloud with Google Cloud Platform
Dr. Ketan Parmar
 
Introduction to Google Cloud Platform
Introduction to Google Cloud PlatformIntroduction to Google Cloud Platform
Introduction to Google Cloud Platform
dhruv_chaudhari
 
Green Cloud Computing
Green Cloud ComputingGreen Cloud Computing
Green Cloud Computing
Seungyun Lee
 

Viewers also liked (18)

Travelling to the far side of Andromeda
Travelling to the far side of AndromedaTravelling to the far side of Andromeda
Travelling to the far side of Andromeda
 
Modeling and Optimization of Resource Allocation in Cloud [PhD Thesis Progres...
Modeling and Optimization of Resource Allocation in Cloud [PhD Thesis Progres...Modeling and Optimization of Resource Allocation in Cloud [PhD Thesis Progres...
Modeling and Optimization of Resource Allocation in Cloud [PhD Thesis Progres...
 
Green cloud computing
Green cloud computingGreen cloud computing
Green cloud computing
 
Plan your Log Platform at Google Cloud Platform
Plan your Log Platform at Google Cloud PlatformPlan your Log Platform at Google Cloud Platform
Plan your Log Platform at Google Cloud Platform
 
A Look Inside Google’s Data Center Networks
A Look Inside Google’s Data Center NetworksA Look Inside Google’s Data Center Networks
A Look Inside Google’s Data Center Networks
 
AI for business: Capire l'opportunità
AI for business: Capire l'opportunitàAI for business: Capire l'opportunità
AI for business: Capire l'opportunità
 
Outsourcing Services: What is right for your organisation
Outsourcing Services: What is right for your organisationOutsourcing Services: What is right for your organisation
Outsourcing Services: What is right for your organisation
 
ONS content extraction
ONS content extractionONS content extraction
ONS content extraction
 
presentationGAATT
presentationGAATTpresentationGAATT
presentationGAATT
 
Serverless Data Architecture at scale on Google Cloud Platform
Serverless Data Architecture at scale on Google Cloud PlatformServerless Data Architecture at scale on Google Cloud Platform
Serverless Data Architecture at scale on Google Cloud Platform
 
Introduction to Google Cloud Platform Technologies
Introduction to Google Cloud Platform TechnologiesIntroduction to Google Cloud Platform Technologies
Introduction to Google Cloud Platform Technologies
 
Green cloud computing
Green cloud computingGreen cloud computing
Green cloud computing
 
Green cloud computing
Green cloud computingGreen cloud computing
Green cloud computing
 
Google Cloud Platform Empowers TensorFlow and Machine Learning
Google Cloud Platform Empowers TensorFlow and Machine LearningGoogle Cloud Platform Empowers TensorFlow and Machine Learning
Google Cloud Platform Empowers TensorFlow and Machine Learning
 
Google Cloud Platform
Google Cloud Platform Google Cloud Platform
Google Cloud Platform
 
Understanding cloud with Google Cloud Platform
Understanding cloud with Google Cloud PlatformUnderstanding cloud with Google Cloud Platform
Understanding cloud with Google Cloud Platform
 
Introduction to Google Cloud Platform
Introduction to Google Cloud PlatformIntroduction to Google Cloud Platform
Introduction to Google Cloud Platform
 
Green Cloud Computing
Green Cloud ComputingGreen Cloud Computing
Green Cloud Computing
 

Similar to Power Comparison Power Comparison of Cloud Data of Cloud Data Center Architectures

MRI Energy-Efficient Cloud Computing
MRI Energy-Efficient Cloud ComputingMRI Energy-Efficient Cloud Computing
MRI Energy-Efficient Cloud ComputingRoger Rafanell Mas
 
HYPPO - NECSTTechTalk 23/04/2020
HYPPO - NECSTTechTalk 23/04/2020HYPPO - NECSTTechTalk 23/04/2020
HYPPO - NECSTTechTalk 23/04/2020
NECST Lab @ Politecnico di Milano
 
DeployingAnAdvancedDistribution.pdf
DeployingAnAdvancedDistribution.pdfDeployingAnAdvancedDistribution.pdf
DeployingAnAdvancedDistribution.pdf
bayu162365
 
The merits of integrating renewables with smarter grid carimet
The merits of integrating renewables with smarter grid   carimetThe merits of integrating renewables with smarter grid   carimet
The merits of integrating renewables with smarter grid carimet
Rick Case, PMP, P.E.
 
Probabilistic consolidation of virtual machines in self organizing cloud data...
Probabilistic consolidation of virtual machines in self organizing cloud data...Probabilistic consolidation of virtual machines in self organizing cloud data...
Probabilistic consolidation of virtual machines in self organizing cloud data...
WMLab,NCU
 
Distributed, concurrent, and independent access to encrypted cloud databases
Distributed, concurrent, and independent access to encrypted cloud databasesDistributed, concurrent, and independent access to encrypted cloud databases
Distributed, concurrent, and independent access to encrypted cloud databases
Papitha Velumani
 
Managing Grid Constraints with Active Management Systems
Managing Grid Constraints with Active Management SystemsManaging Grid Constraints with Active Management Systems
Managing Grid Constraints with Active Management Systems
Smarter Grid Solutions
 
10.6_Utility Microgrids_Reid_EPRI/SNL Microgrid Symposium
10.6_Utility Microgrids_Reid_EPRI/SNL Microgrid Symposium10.6_Utility Microgrids_Reid_EPRI/SNL Microgrid Symposium
10.6_Utility Microgrids_Reid_EPRI/SNL Microgrid Symposium
Sandia National Laboratories: Energy & Climate: Renewables
 
Dynamic workload migration over optical backbone network to minimize data cen...
Dynamic workload migration over optical backbone network to minimize data cen...Dynamic workload migration over optical backbone network to minimize data cen...
Dynamic workload migration over optical backbone network to minimize data cen...
Sabidur Rahman
 
2.4_Overview of Microgrid Research, Development, and Resiliency Analysis_Hovs...
2.4_Overview of Microgrid Research, Development, and Resiliency Analysis_Hovs...2.4_Overview of Microgrid Research, Development, and Resiliency Analysis_Hovs...
2.4_Overview of Microgrid Research, Development, and Resiliency Analysis_Hovs...
Sandia National Laboratories: Energy & Climate: Renewables
 
Green ICT
Green ICTGreen ICT
Green ICT
Sähköklubi
 
Distributed, concurrent, and independent access to encrypted cloud databases
Distributed, concurrent, and independent access to encrypted cloud databasesDistributed, concurrent, and independent access to encrypted cloud databases
Distributed, concurrent, and independent access to encrypted cloud databases
Papitha Velumani
 
Distributed, concurrent, and independent access to encrypted cloud databases
Distributed, concurrent, and independent access to encrypted cloud databasesDistributed, concurrent, and independent access to encrypted cloud databases
Distributed, concurrent, and independent access to encrypted cloud databases
Papitha Velumani
 
Koomeyondatacenterelectricityuse v24
Koomeyondatacenterelectricityuse v24Koomeyondatacenterelectricityuse v24
Koomeyondatacenterelectricityuse v24
Jonathan Koomey
 
The electricity distribution system and energy decarbonisation
The electricity distribution system and energy decarbonisationThe electricity distribution system and energy decarbonisation
The electricity distribution system and energy decarbonisation
Economic and Social Research Institute
 
T1-4_Maslennikov_et_al.pdf
T1-4_Maslennikov_et_al.pdfT1-4_Maslennikov_et_al.pdf
T1-4_Maslennikov_et_al.pdf
MareLunare
 
Metering Energy Consumption in Data Centres - Colin Love
Metering Energy Consumption in Data Centres - Colin LoveMetering Energy Consumption in Data Centres - Colin Love
Metering Energy Consumption in Data Centres - Colin LoveGoodCampus
 
Hacking the Conventional Data Center Infrastructure
Hacking the Conventional  Data Center InfrastructureHacking the Conventional  Data Center Infrastructure
Hacking the Conventional Data Center Infrastructure
Schneider Electric
 
Energy efficient computing & computational services
Energy efficient computing & computational services Energy efficient computing & computational services
Energy efficient computing & computational services
David Wallom
 
CNR @ VMUG.IT 20150304
CNR @ VMUG.IT 20150304CNR @ VMUG.IT 20150304
CNR @ VMUG.IT 20150304
VMUG IT
 

Similar to Power Comparison Power Comparison of Cloud Data of Cloud Data Center Architectures (20)

MRI Energy-Efficient Cloud Computing
MRI Energy-Efficient Cloud ComputingMRI Energy-Efficient Cloud Computing
MRI Energy-Efficient Cloud Computing
 
HYPPO - NECSTTechTalk 23/04/2020
HYPPO - NECSTTechTalk 23/04/2020HYPPO - NECSTTechTalk 23/04/2020
HYPPO - NECSTTechTalk 23/04/2020
 
DeployingAnAdvancedDistribution.pdf
DeployingAnAdvancedDistribution.pdfDeployingAnAdvancedDistribution.pdf
DeployingAnAdvancedDistribution.pdf
 
The merits of integrating renewables with smarter grid carimet
The merits of integrating renewables with smarter grid   carimetThe merits of integrating renewables with smarter grid   carimet
The merits of integrating renewables with smarter grid carimet
 
Probabilistic consolidation of virtual machines in self organizing cloud data...
Probabilistic consolidation of virtual machines in self organizing cloud data...Probabilistic consolidation of virtual machines in self organizing cloud data...
Probabilistic consolidation of virtual machines in self organizing cloud data...
 
Distributed, concurrent, and independent access to encrypted cloud databases
Distributed, concurrent, and independent access to encrypted cloud databasesDistributed, concurrent, and independent access to encrypted cloud databases
Distributed, concurrent, and independent access to encrypted cloud databases
 
Managing Grid Constraints with Active Management Systems
Managing Grid Constraints with Active Management SystemsManaging Grid Constraints with Active Management Systems
Managing Grid Constraints with Active Management Systems
 
10.6_Utility Microgrids_Reid_EPRI/SNL Microgrid Symposium
10.6_Utility Microgrids_Reid_EPRI/SNL Microgrid Symposium10.6_Utility Microgrids_Reid_EPRI/SNL Microgrid Symposium
10.6_Utility Microgrids_Reid_EPRI/SNL Microgrid Symposium
 
Dynamic workload migration over optical backbone network to minimize data cen...
Dynamic workload migration over optical backbone network to minimize data cen...Dynamic workload migration over optical backbone network to minimize data cen...
Dynamic workload migration over optical backbone network to minimize data cen...
 
2.4_Overview of Microgrid Research, Development, and Resiliency Analysis_Hovs...
2.4_Overview of Microgrid Research, Development, and Resiliency Analysis_Hovs...2.4_Overview of Microgrid Research, Development, and Resiliency Analysis_Hovs...
2.4_Overview of Microgrid Research, Development, and Resiliency Analysis_Hovs...
 
Green ICT
Green ICTGreen ICT
Green ICT
 
Distributed, concurrent, and independent access to encrypted cloud databases
Distributed, concurrent, and independent access to encrypted cloud databasesDistributed, concurrent, and independent access to encrypted cloud databases
Distributed, concurrent, and independent access to encrypted cloud databases
 
Distributed, concurrent, and independent access to encrypted cloud databases
Distributed, concurrent, and independent access to encrypted cloud databasesDistributed, concurrent, and independent access to encrypted cloud databases
Distributed, concurrent, and independent access to encrypted cloud databases
 
Koomeyondatacenterelectricityuse v24
Koomeyondatacenterelectricityuse v24Koomeyondatacenterelectricityuse v24
Koomeyondatacenterelectricityuse v24
 
The electricity distribution system and energy decarbonisation
The electricity distribution system and energy decarbonisationThe electricity distribution system and energy decarbonisation
The electricity distribution system and energy decarbonisation
 
T1-4_Maslennikov_et_al.pdf
T1-4_Maslennikov_et_al.pdfT1-4_Maslennikov_et_al.pdf
T1-4_Maslennikov_et_al.pdf
 
Metering Energy Consumption in Data Centres - Colin Love
Metering Energy Consumption in Data Centres - Colin LoveMetering Energy Consumption in Data Centres - Colin Love
Metering Energy Consumption in Data Centres - Colin Love
 
Hacking the Conventional Data Center Infrastructure
Hacking the Conventional  Data Center InfrastructureHacking the Conventional  Data Center Infrastructure
Hacking the Conventional Data Center Infrastructure
 
Energy efficient computing & computational services
Energy efficient computing & computational services Energy efficient computing & computational services
Energy efficient computing & computational services
 
CNR @ VMUG.IT 20150304
CNR @ VMUG.IT 20150304CNR @ VMUG.IT 20150304
CNR @ VMUG.IT 20150304
 

Recently uploaded

KuberTENes Birthday Bash Guadalajara - K8sGPT first impressions
KuberTENes Birthday Bash Guadalajara - K8sGPT first impressionsKuberTENes Birthday Bash Guadalajara - K8sGPT first impressions
KuberTENes Birthday Bash Guadalajara - K8sGPT first impressions
Victor Morales
 
spirit beverages ppt without graphics.pptx
spirit beverages ppt without graphics.pptxspirit beverages ppt without graphics.pptx
spirit beverages ppt without graphics.pptx
Madan Karki
 
CHINA’S GEO-ECONOMIC OUTREACH IN CENTRAL ASIAN COUNTRIES AND FUTURE PROSPECT
CHINA’S GEO-ECONOMIC OUTREACH IN CENTRAL ASIAN COUNTRIES AND FUTURE PROSPECTCHINA’S GEO-ECONOMIC OUTREACH IN CENTRAL ASIAN COUNTRIES AND FUTURE PROSPECT
CHINA’S GEO-ECONOMIC OUTREACH IN CENTRAL ASIAN COUNTRIES AND FUTURE PROSPECT
jpsjournal1
 
5214-1693458878915-Unit 6 2023 to 2024 academic year assignment (AutoRecovere...
5214-1693458878915-Unit 6 2023 to 2024 academic year assignment (AutoRecovere...5214-1693458878915-Unit 6 2023 to 2024 academic year assignment (AutoRecovere...
5214-1693458878915-Unit 6 2023 to 2024 academic year assignment (AutoRecovere...
ihlasbinance2003
 
PROJECT FORMAT FOR EVS AMITY UNIVERSITY GWALIOR.ppt
PROJECT FORMAT FOR EVS AMITY UNIVERSITY GWALIOR.pptPROJECT FORMAT FOR EVS AMITY UNIVERSITY GWALIOR.ppt
PROJECT FORMAT FOR EVS AMITY UNIVERSITY GWALIOR.ppt
bhadouriyakaku
 
PPT on GRP pipes manufacturing and testing
PPT on GRP pipes manufacturing and testingPPT on GRP pipes manufacturing and testing
PPT on GRP pipes manufacturing and testing
anoopmanoharan2
 
introduction to solar energy for engineering.pdf
introduction to solar energy for engineering.pdfintroduction to solar energy for engineering.pdf
introduction to solar energy for engineering.pdf
ravindarpurohit26
 
Hierarchical Digital Twin of a Naval Power System
Hierarchical Digital Twin of a Naval Power SystemHierarchical Digital Twin of a Naval Power System
Hierarchical Digital Twin of a Naval Power System
Kerry Sado
 
DEEP LEARNING FOR SMART GRID INTRUSION DETECTION: A HYBRID CNN-LSTM-BASED MODEL
DEEP LEARNING FOR SMART GRID INTRUSION DETECTION: A HYBRID CNN-LSTM-BASED MODELDEEP LEARNING FOR SMART GRID INTRUSION DETECTION: A HYBRID CNN-LSTM-BASED MODEL
DEEP LEARNING FOR SMART GRID INTRUSION DETECTION: A HYBRID CNN-LSTM-BASED MODEL
gerogepatton
 
Unbalanced Three Phase Systems and circuits.pptx
Unbalanced Three Phase Systems and circuits.pptxUnbalanced Three Phase Systems and circuits.pptx
Unbalanced Three Phase Systems and circuits.pptx
ChristineTorrepenida1
 
ACRP 4-09 Risk Assessment Method to Support Modification of Airfield Separat...
ACRP 4-09 Risk Assessment Method to Support Modification of Airfield Separat...ACRP 4-09 Risk Assessment Method to Support Modification of Airfield Separat...
ACRP 4-09 Risk Assessment Method to Support Modification of Airfield Separat...
Mukeshwaran Balu
 
Series of visio cisco devices Cisco_Icons.ppt
Series of visio cisco devices Cisco_Icons.pptSeries of visio cisco devices Cisco_Icons.ppt
Series of visio cisco devices Cisco_Icons.ppt
PauloRodrigues104553
 
Harnessing WebAssembly for Real-time Stateless Streaming Pipelines
Harnessing WebAssembly for Real-time Stateless Streaming PipelinesHarnessing WebAssembly for Real-time Stateless Streaming Pipelines
Harnessing WebAssembly for Real-time Stateless Streaming Pipelines
Christina Lin
 
Technical Drawings introduction to drawing of prisms
Technical Drawings introduction to drawing of prismsTechnical Drawings introduction to drawing of prisms
Technical Drawings introduction to drawing of prisms
heavyhaig
 
哪里办理(csu毕业证书)查尔斯特大学毕业证硕士学历原版一模一样
哪里办理(csu毕业证书)查尔斯特大学毕业证硕士学历原版一模一样哪里办理(csu毕业证书)查尔斯特大学毕业证硕士学历原版一模一样
哪里办理(csu毕业证书)查尔斯特大学毕业证硕士学历原版一模一样
insn4465
 
Literature Review Basics and Understanding Reference Management.pptx
Literature Review Basics and Understanding Reference Management.pptxLiterature Review Basics and Understanding Reference Management.pptx
Literature Review Basics and Understanding Reference Management.pptx
Dr Ramhari Poudyal
 
New techniques for characterising damage in rock slopes.pdf
New techniques for characterising damage in rock slopes.pdfNew techniques for characterising damage in rock slopes.pdf
New techniques for characterising damage in rock slopes.pdf
wisnuprabawa3
 
sieving analysis and results interpretation
sieving analysis and results interpretationsieving analysis and results interpretation
sieving analysis and results interpretation
ssuser36d3051
 
Design and Analysis of Algorithms-DP,Backtracking,Graphs,B&B
Design and Analysis of Algorithms-DP,Backtracking,Graphs,B&BDesign and Analysis of Algorithms-DP,Backtracking,Graphs,B&B
Design and Analysis of Algorithms-DP,Backtracking,Graphs,B&B
Sreedhar Chowdam
 
Building Electrical System Design & Installation
Building Electrical System Design & InstallationBuilding Electrical System Design & Installation
Building Electrical System Design & Installation
symbo111
 

Recently uploaded (20)

KuberTENes Birthday Bash Guadalajara - K8sGPT first impressions
KuberTENes Birthday Bash Guadalajara - K8sGPT first impressionsKuberTENes Birthday Bash Guadalajara - K8sGPT first impressions
KuberTENes Birthday Bash Guadalajara - K8sGPT first impressions
 
spirit beverages ppt without graphics.pptx
spirit beverages ppt without graphics.pptxspirit beverages ppt without graphics.pptx
spirit beverages ppt without graphics.pptx
 
CHINA’S GEO-ECONOMIC OUTREACH IN CENTRAL ASIAN COUNTRIES AND FUTURE PROSPECT
CHINA’S GEO-ECONOMIC OUTREACH IN CENTRAL ASIAN COUNTRIES AND FUTURE PROSPECTCHINA’S GEO-ECONOMIC OUTREACH IN CENTRAL ASIAN COUNTRIES AND FUTURE PROSPECT
CHINA’S GEO-ECONOMIC OUTREACH IN CENTRAL ASIAN COUNTRIES AND FUTURE PROSPECT
 
5214-1693458878915-Unit 6 2023 to 2024 academic year assignment (AutoRecovere...
5214-1693458878915-Unit 6 2023 to 2024 academic year assignment (AutoRecovere...5214-1693458878915-Unit 6 2023 to 2024 academic year assignment (AutoRecovere...
5214-1693458878915-Unit 6 2023 to 2024 academic year assignment (AutoRecovere...
 
PROJECT FORMAT FOR EVS AMITY UNIVERSITY GWALIOR.ppt
PROJECT FORMAT FOR EVS AMITY UNIVERSITY GWALIOR.pptPROJECT FORMAT FOR EVS AMITY UNIVERSITY GWALIOR.ppt
PROJECT FORMAT FOR EVS AMITY UNIVERSITY GWALIOR.ppt
 
PPT on GRP pipes manufacturing and testing
PPT on GRP pipes manufacturing and testingPPT on GRP pipes manufacturing and testing
PPT on GRP pipes manufacturing and testing
 
introduction to solar energy for engineering.pdf
introduction to solar energy for engineering.pdfintroduction to solar energy for engineering.pdf
introduction to solar energy for engineering.pdf
 
Hierarchical Digital Twin of a Naval Power System
Hierarchical Digital Twin of a Naval Power SystemHierarchical Digital Twin of a Naval Power System
Hierarchical Digital Twin of a Naval Power System
 
DEEP LEARNING FOR SMART GRID INTRUSION DETECTION: A HYBRID CNN-LSTM-BASED MODEL
DEEP LEARNING FOR SMART GRID INTRUSION DETECTION: A HYBRID CNN-LSTM-BASED MODELDEEP LEARNING FOR SMART GRID INTRUSION DETECTION: A HYBRID CNN-LSTM-BASED MODEL
DEEP LEARNING FOR SMART GRID INTRUSION DETECTION: A HYBRID CNN-LSTM-BASED MODEL
 
Unbalanced Three Phase Systems and circuits.pptx
Unbalanced Three Phase Systems and circuits.pptxUnbalanced Three Phase Systems and circuits.pptx
Unbalanced Three Phase Systems and circuits.pptx
 
ACRP 4-09 Risk Assessment Method to Support Modification of Airfield Separat...
ACRP 4-09 Risk Assessment Method to Support Modification of Airfield Separat...ACRP 4-09 Risk Assessment Method to Support Modification of Airfield Separat...
ACRP 4-09 Risk Assessment Method to Support Modification of Airfield Separat...
 
Series of visio cisco devices Cisco_Icons.ppt
Series of visio cisco devices Cisco_Icons.pptSeries of visio cisco devices Cisco_Icons.ppt
Series of visio cisco devices Cisco_Icons.ppt
 
Harnessing WebAssembly for Real-time Stateless Streaming Pipelines
Harnessing WebAssembly for Real-time Stateless Streaming PipelinesHarnessing WebAssembly for Real-time Stateless Streaming Pipelines
Harnessing WebAssembly for Real-time Stateless Streaming Pipelines
 
Technical Drawings introduction to drawing of prisms
Technical Drawings introduction to drawing of prismsTechnical Drawings introduction to drawing of prisms
Technical Drawings introduction to drawing of prisms
 
哪里办理(csu毕业证书)查尔斯特大学毕业证硕士学历原版一模一样
哪里办理(csu毕业证书)查尔斯特大学毕业证硕士学历原版一模一样哪里办理(csu毕业证书)查尔斯特大学毕业证硕士学历原版一模一样
哪里办理(csu毕业证书)查尔斯特大学毕业证硕士学历原版一模一样
 
Literature Review Basics and Understanding Reference Management.pptx
Literature Review Basics and Understanding Reference Management.pptxLiterature Review Basics and Understanding Reference Management.pptx
Literature Review Basics and Understanding Reference Management.pptx
 
New techniques for characterising damage in rock slopes.pdf
New techniques for characterising damage in rock slopes.pdfNew techniques for characterising damage in rock slopes.pdf
New techniques for characterising damage in rock slopes.pdf
 
sieving analysis and results interpretation
sieving analysis and results interpretationsieving analysis and results interpretation
sieving analysis and results interpretation
 
Design and Analysis of Algorithms-DP,Backtracking,Graphs,B&B
Design and Analysis of Algorithms-DP,Backtracking,Graphs,B&BDesign and Analysis of Algorithms-DP,Backtracking,Graphs,B&B
Design and Analysis of Algorithms-DP,Backtracking,Graphs,B&B
 
Building Electrical System Design & Installation
Building Electrical System Design & InstallationBuilding Electrical System Design & Installation
Building Electrical System Design & Installation
 

Power Comparison Power Comparison of Cloud Data of Cloud Data Center Architectures

  • 1. Power Comparison of Cloud Data Center Architectures Pietro Ruiu Andrea Bianco, Paolo Giaccone Claudio Fiandrino, Dzmitry Kliazovich 13th Italian Networking Workshop: San Candido, Italy January 13 - 15, 2016
  • 2. The quest for green data centers • Data centers are the new polluters of 21st century – in 2012, accounted for 15% of the global ICT energy consumption – expected to increase in the next years • Data center consumption – 75% ICT equipment • powering and cooling • mostly due to servers – 25% power distribution and facility operations • Strong interest in designing and operating data centers with higher energy efficiency – not only to reduce OPEX costs 2
  • 3. Key question • Given • a data center topology • a power consumption profile for each ICT device • Define the min-power job allocation policy • Evaluate power consumption as function of the data center load Classical questions • Given • a power consumption profile for each ICT device • a generic job allocation policy • Compare the power consumption behavior in function of the data center topology Our question 3
  • 4. Data center model 4 Data center network (DCN) Power Load Power Load Power Load Servers ICT device Power profile
  • 5. Local vs global energy proportionality • consume proportional to the load • consume (and pay) only if really needed Ideal energy proportionality • Constant power (CONST) • Full Energy Proportional (FEP) • Linear (LIN) Local power consumption for single device • maybe very different from local power consumption Global power consumption for overall system 5 CONST Power Load FEP Power Load LIN Power Load Power Load
  • 6. Global power consumption 6 • N resources/devices to be allocated for a set of requests/VMs • power consumption profile for each resource/device • allocation policy • consolidate: activate the minimum number of resources • load-balance: distribute load across the resources Resource allocation • depends on granularity of the resources (i.e. the value of N) • depends on allocation policy Overall power consumption
  • 7. Global Power Consumption 7 Local power consumption Consolidate policy Load-balance policy Normalized power = Power / Load FEP Power Load Power Load Load Norm.Power Power Load Load Norm.Power CONST Power Load N Load Power N Load Power N Load Norm.Power N Load Norm.Power • Local energy proportionality implies global energy proportionality • If N is enough large, consolidate policy reaches global energy proportionality for CONST local power Local and Global Energy Proportionality
  • 8. Our contributions • Network-aware min-power VM online allocation policy • Flow-level C++ simulator of data centers • Power comparison of different network topologies 8 • Energy profiles for each ICT device (switch, link, server) • DCN topology • VM arrival process Flow-level simulator • Global power consumption • Load on each ICT device • VM blocking probabilityVM allocation policy
  • 9. DCN topologies • traditional Clos-based switch topologies – classical 2, 3 tiers – Jupiter • Google’s disclosed DCN architecture • “Jupiter Rising: A Decade of Clos Topologies and Centralized Control in Google’s Datacenter Network”, ACM SIGCOMM CCR, Oct. 2015 9
  • 10. • Switches: 10 core sw – 18 TOR sw • Servers 180 • Total ICT devices: 208 nodes CORE TOR 10Gbps40Gbps 2-tiers DCN 10
  • 11. • 3 core sw – 6 aggregation sw – 18 TOR sw • 180 servers – 27 switches – 207 nodes CORE AGGREGATION TOR 10Gbps40Gbps40Gbps 3-tiers DCN 11
  • 12. • 24 spine sw – 16 aggregation sw – 16 TOR sw • 192 servers – 44 switches – 236 nodes = 4p@40Gbps (or 16p@10Gbps) SPINEAGGREGATIONTOR 10Gbps10Gbps40Gbps40Gbps40Gbps MB MB MB MB Jupiter-like DCN 12
  • 13. Online VM allocation policy 13 • for each VM, select a server at random • connect through the minimum incremental DCN power • load-balance on the servers RSS (Random Server Selection) • for each VM, select the server with minimum incremental power (server + DCN) • consolidate VMs in the same server, in the same rack, in closeby racks, etc • variant of min-cost Dijkstra algorithm MNP (Minimum Network Power)
  • 14. VM generation • time is slotted • at each timelot, a new VM arrives and must communicate B bps to a previously randomly allocated VM – B is randomly chosen – destination VM is chosen with Bernoulli trials • simulation can run until saturating the data center 14 VM1 VM2 VM3 VM4 VM5
  • 15. RSS (Random Server Selection) • small datacenter (180-192 servers) • Jupiter appears to be the most energy proportional – due to the larger number of switches (44 vs 27-28) 15 0 5 10 15 20 25 20 30 40 50 60 70 80 90 100 NormalizedpowerperVM[W] Data Center load [%] 3-Tier LIN FEP CONST 0 5 10 15 20 25 20 30 40 50 60 70 80 90 100 Data Center load [%] 2-Tier LIN FEP CONST 0 5 10 15 20 25 20 30 40 50 60 70 80 90 100 Data Center load [%] Jupiter LIN FEP CONST
  • 16. MNP (Minimum Network Power) 16 • small datacenter (180-192 servers) • MNP allows to achieve global energy-proportionality • under FEP, power jumps due to abrupt activation of new layers 0 0.5 1 1.5 2 2.5 3 20 30 40 50 60 70 80 90100 NormalizedpowerperVM[W] Data Center load [%] 3-Tier LIN FEP CONST 0 0.5 1 1.5 2 2.5 3 20 30 40 50 60 70 80 90 100 Data Center load [%] 2-Tier LIN FEP CONST 0 0.5 1 1.5 2 2.5 3 20 30 40 50 60 70 80 90 100 Data Center load [%] Jupiter LIN FEP CONST
  • 17. Conclusions • Global energy proportionality of an overall data center depends on • local power profile of each device • topology (number of devices) • VM allocation policy Take-home message • consider large topologies with 10,000 servers • compare data center networks given the same bisection bandwidth • consider the allocation of clusters of VMs Future works 17

Editor's Notes

  1. Perche’ LIN si comporta cosi’ male???