SlideShare une entreprise Scribd logo
1  sur  1
Télécharger pour lire hors ligne
TEMPLATE DESIGN © 2008
www.PosterPresentations.com
Auto-Scaling Frameworks for Big Data Analytics on Clouds
 The primary goal of big data analytics is to help companies make more
informed business decisions.
 Uncover hidden patterns and unknown correlations between data.
 High-performance analytics necessary to determine the relevance of data.
 Gartner states that through 2015, 85% of Fortune 500 organizations will
be unable to exploit big data for competitive advantage. [1]
 Traditional data processing tools cannot handle a large volume of data.
 Google’s MapReduce programming model is able to handle big data.
 MapReduce requires multiple compute resources to process large tasks.
 The demand to process data varies based on the current period.
 A MapReduce model can be used together with Cloud Computing.
 Cloud computing enables companies to consume compute resources as a
utility rather building and maintaining computing infrastructures in-house.
 To handle variable load while processing data, a technique known as
Auto-Scaling is employed.
 An Intermediary enterprise handles user
jobs from a single client enterprise and
executes those jobs on resources acquired
from a public cloud provider to form a
virtual private cloud for the users.
 The broker is hosted in the intermediary
enterprise.
 The public cloud provider charges the
broker c_pub dollars an hour per
resource.
 The broker charges the user c_pvt dollars
per second.
 A novel proactive [3] and a reactive [4] auto-scaling framework which uses
a price model that can lead to an increase in the profit for the broker
(intermediary enterprise) and a reduction in the user cost at the same
time.
 Both frameworks handle SLA-driven advanced reservation requests as
well as well as On-Demand requests.
 A detailed performance analysis focusing on broker profit and user
cost for a prototype subjected to various combinations of system and
workload parameters is performed and key insights into system behavior
is presented.
 System II: A system that can scale up if it can not fit a request on the
resources available.
 Resources are scaled down when they are no longer needed.
 Comparing broker profit and total user cost of System II with System I.
 System III: The users purchase resources directly from Amazon.
 There is no broker present in this system.
 Comparing total user cost of System III with System I and System II.
 Auto-Scaling is the ability to modify the capacity for the user’s cloud
infrastructure based on traffic patterns.
 Benefits of Auto-Scaling:
 Fault tolerant
 Highly available
 No Capacity Planning
 Amazon’s CloudWatch allow users to set thresholds for Auto-Scaling.
 Qubole [2] offer cloud based data analytics services based on the
MapReduce programming Model with Apache’s Hadoop and Hive.
 Qubole uses resources from the Amazon Cloud and allows auto-scaling
based on user’s workload.
 Qubole offers cost savings by optimizing data processing tasks.
 However, these services do not handle deadlines with individual user
jobs which perform the data analytics tasks.
Two types of Auto-Scaling techniques are discussed:
 A proactive framework that scales resources based on predictions from
a machine learning engine (MLE). This system scales independent off the
user demand.
 A reactive framework that scales resources when a request arrives to
the system. This system scales depending on the user demand.
Effect of Load Factor on BP and UC
 Load Factor (f) – is the ratio of the number of requests generated with
an arrival rate of λlow to the total number of requests generated during
the experiment.
Performance metrics:
 Broker profit (BP) - is the profit a broker earns ($/hour )
 Total User cost (UC) - is the amount charged to the user ($/hour)
 The number of resources in the pool used by the user requests need not
be determined a priori and are controlled dynamically thereby reducing
the cost for capacity planning.
 Overall, System I preforms better than System II and System III in terms
of Broker Profit and User Cost.
 The Auto-Scaling frameworks accept a MapReduce workload that
contain jobs to analyze big data. Other types of workloads may also be
handled by the Auto-Scaling system.
 Both frameworks allow users in an enterprise for example, to submit
Advance Reservation as well as On-Demand requests to a private cloud
provided by the intermediary cloud provider.
 The reactive framework provides a minimum grade of service to ensure
that a certain percentage of user requests are guaranteed to be
executed.
Introduction 1 Contributions 5
The Broker Model
3
Other Systems for Performance Evaluations 7
Conclusions 9
This research is supported by the Natural Sciences and Engineering
Council of Canada (NSERC) and TELUS
----------------------------------------------------------------------------------------------------
[1] Gartner Report, Big Data Management & Analytics, http://www.gartner.com/technology/topics/big-data.jsp.
[2] Qubole, http://www.qubole.com/, Accessed March 2015.
[3] A. Biswas, S. Majumdar, B. Nandy and A. El-Haraki, "Automatic Resource Provisioning: a Machine Learning based Proactive
approach," in Proc. of International Conference on Cloud Computing Technology and Science (CloudCom), Singapore, 12 2014.
[4] A. Biswas, S. Majumdar, B. Nandy and A. El-Haraki, "An Auto-scaling Framework for Controlling Enterprise Resources on Clouds," in
Proc. of the 15th IEEE/ACM International Symposium on Cluster, Cloud and Grid Computing (CCGrid), Shenzhen, 05, 2015.
Acknowledgements and References 11
Auto-Scaling 2
Auto-scaling decisions can be made:
Reactively: the system reacts to changes in user workload.
 Scaling conditions based on a target metric reaching some threshold.
 Set a threshold for a metric which triggers an adjustment of resources.
Proactively: the system tries to predict future resource requirements in
order to ensure sufficient resource is available ahead of time.
 Based on time series analysis or control theory.
 Predict the workload based on past demand.
Types of Auto-Scaling 3
http://docs.aws.amazon.com/AutoScaling/latest/DeveloperGuide/
WhatIsAutoScaling.html Section “What is Auto Scaling”
4
Internet
Single Client Enterprise
Broker Virtual Private Cloud
Resources
End Users
Public Cloud
Provider
Intermediary
Cloud Provider
Auto-Scaling frameworks 6
Sample Performance Results (Proactive System) 8
 BP decreases with f. System I earns more profit that System II.
 UC decreases with f. System I charges users a lesser amount that
System II and System III.
 BP increases with increase in Service Time and Laxity Factor and
System I earns more profit that System II.
Effect of Service Time and Laxity on BP
Anshuman Biswas
Systems and Computer Engineering
Carleton University
anshuman@sce.carleton.ca
Supervisors: Shikharesh Majumdar, Biswajit Nandy
Systems and Computer Engineering
Carleton University
{majumdar,bnandy}@sce.carleton.ca
Collaborator:
Ali El-Haraki
TELUS
User requests have Service
Level Agreements (SLAs):
Arrival Time, Earliest start
time , Execution time and
Deadline
RequestHandler
User
Request
MatchMake
Sched
Reactive
Autoscaler
Dynamic
Resouce
Pool
Manager
Decision
Maker
GoS
Decision Maker
Predictive
Autoscaler
Dynamic
Resource
Pool
Manager
RequestHandler
Machine Learning
Engine
MatchMake
Sched
User Request
Reactive Auto-Scaler
Proactive Auto-Scaler
The Reactive Auto-Scaler:
 Request Handler (RH) is responsible for handling user requests.
 RH forwards this request to Decision Maker (DM) which decided whether
to accept or reject the request based on two operations.
DM sends the request to the Matchmaking and scheduling component (MMS)
which is responsible for matchmaking and scheduling the request.
If MMS cannot schedule the request, DM then invokes the reactive auto-scaler.
 A criterion for the reactive Auto-Scaler for acquiring resources is based on
grade of service (GoS).
The GoS is specified by the client enterprise.
The system described uses the blocking ratio (B) as the GoS.
B is the proportion of requests that cannot be completed before the expiry of
their deadlines and are therefore rejected by the system.
 For the reactive Auto-Scaler, DM consults with a GoS.
 Framework provides a minimum Quality of Service (QoS) for the users.
 Acquiring and releasing a resource is based on these three rules:
Rule I: When B > Bspec acquire resource, where Bspec is the desired value of B
(GoS) maintained by the broker.
 Rule II: When (Broker Profit)i > 0, acquire ith resource.
Rule III: When (stopi) = current time, release ith resource, where stopi is the
stop time for the ith resource.
The Proactive Auto-Scaler:
 This broker replaces the GoS module with the Machine Learning Engine
(MLE) Module.
 MLE predicts the requests arriving in the future and then auto-scales.
 Auto-scaling resources is based on Rule II and Rule III from the Reactive broker.
 MLE predicts after a set number of requests have arrived in the system.
The reactive/proactive auto-scaling system is known as System I.
Future Work 10
Detailed performance analysis using reactive auto-scaling technique.
Additional simulation experiments for various other combination of system
and workload parameters are being planned.
Using traces of real workload in simulation can lead to interesting insights
into system behaviour and performance.

Contenu connexe

Tendances

AUTO RESOURCE MANAGEMENT TO ENHANCE RELIABILITY AND ENERGY CONSUMPTION IN HET...
AUTO RESOURCE MANAGEMENT TO ENHANCE RELIABILITY AND ENERGY CONSUMPTION IN HET...AUTO RESOURCE MANAGEMENT TO ENHANCE RELIABILITY AND ENERGY CONSUMPTION IN HET...
AUTO RESOURCE MANAGEMENT TO ENHANCE RELIABILITY AND ENERGY CONSUMPTION IN HET...IJCNCJournal
 
A latency-aware max-min algorithm for resource allocation in cloud
A latency-aware max-min algorithm for resource  allocation in cloud A latency-aware max-min algorithm for resource  allocation in cloud
A latency-aware max-min algorithm for resource allocation in cloud IJECEIAES
 
Iaetsd effective fault toerant resource allocation with cost
Iaetsd effective fault toerant resource allocation with costIaetsd effective fault toerant resource allocation with cost
Iaetsd effective fault toerant resource allocation with costIaetsd Iaetsd
 
Mod05lec22(cloudonomics tutorial)
Mod05lec22(cloudonomics tutorial)Mod05lec22(cloudonomics tutorial)
Mod05lec22(cloudonomics tutorial)Ankit Gupta
 
Survey on Dynamic Resource Allocation Strategy in Cloud Computing Environment
Survey on Dynamic Resource Allocation Strategy in Cloud Computing EnvironmentSurvey on Dynamic Resource Allocation Strategy in Cloud Computing Environment
Survey on Dynamic Resource Allocation Strategy in Cloud Computing EnvironmentEditor IJCATR
 
Allocation Strategies of Virtual Resources in Cloud-Computing Networks
Allocation Strategies of Virtual Resources in Cloud-Computing NetworksAllocation Strategies of Virtual Resources in Cloud-Computing Networks
Allocation Strategies of Virtual Resources in Cloud-Computing NetworksIJERA Editor
 
International Journal of Engineering Research and Development
International Journal of Engineering Research and DevelopmentInternational Journal of Engineering Research and Development
International Journal of Engineering Research and DevelopmentIJERD Editor
 
B02120307013
B02120307013B02120307013
B02120307013theijes
 
A trust management system for ad hoc mobile
A trust management system for ad hoc mobileA trust management system for ad hoc mobile
A trust management system for ad hoc mobileAhmed Hammam
 
Hybrid Based Resource Provisioning in Cloud
Hybrid Based Resource Provisioning in CloudHybrid Based Resource Provisioning in Cloud
Hybrid Based Resource Provisioning in CloudEditor IJCATR
 
Extending Grids with Cloud Resource Management for Scientific Computing
Extending Grids with Cloud Resource Management for Scientific ComputingExtending Grids with Cloud Resource Management for Scientific Computing
Extending Grids with Cloud Resource Management for Scientific ComputingBharat Kalia
 
International Refereed Journal of Engineering and Science (IRJES)
International Refereed Journal of Engineering and Science (IRJES)International Refereed Journal of Engineering and Science (IRJES)
International Refereed Journal of Engineering and Science (IRJES)irjes
 
A REVIEW ON LOAD BALANCING IN CLOUD USING ENHANCED GENETIC ALGORITHM
A REVIEW ON LOAD BALANCING IN CLOUD USING ENHANCED GENETIC ALGORITHM A REVIEW ON LOAD BALANCING IN CLOUD USING ENHANCED GENETIC ALGORITHM
A REVIEW ON LOAD BALANCING IN CLOUD USING ENHANCED GENETIC ALGORITHM IAEME Publication
 

Tendances (15)

C017531925
C017531925C017531925
C017531925
 
AUTO RESOURCE MANAGEMENT TO ENHANCE RELIABILITY AND ENERGY CONSUMPTION IN HET...
AUTO RESOURCE MANAGEMENT TO ENHANCE RELIABILITY AND ENERGY CONSUMPTION IN HET...AUTO RESOURCE MANAGEMENT TO ENHANCE RELIABILITY AND ENERGY CONSUMPTION IN HET...
AUTO RESOURCE MANAGEMENT TO ENHANCE RELIABILITY AND ENERGY CONSUMPTION IN HET...
 
A latency-aware max-min algorithm for resource allocation in cloud
A latency-aware max-min algorithm for resource  allocation in cloud A latency-aware max-min algorithm for resource  allocation in cloud
A latency-aware max-min algorithm for resource allocation in cloud
 
Iaetsd effective fault toerant resource allocation with cost
Iaetsd effective fault toerant resource allocation with costIaetsd effective fault toerant resource allocation with cost
Iaetsd effective fault toerant resource allocation with cost
 
Mod05lec22(cloudonomics tutorial)
Mod05lec22(cloudonomics tutorial)Mod05lec22(cloudonomics tutorial)
Mod05lec22(cloudonomics tutorial)
 
Survey on Dynamic Resource Allocation Strategy in Cloud Computing Environment
Survey on Dynamic Resource Allocation Strategy in Cloud Computing EnvironmentSurvey on Dynamic Resource Allocation Strategy in Cloud Computing Environment
Survey on Dynamic Resource Allocation Strategy in Cloud Computing Environment
 
Allocation Strategies of Virtual Resources in Cloud-Computing Networks
Allocation Strategies of Virtual Resources in Cloud-Computing NetworksAllocation Strategies of Virtual Resources in Cloud-Computing Networks
Allocation Strategies of Virtual Resources in Cloud-Computing Networks
 
T04503113118
T04503113118T04503113118
T04503113118
 
International Journal of Engineering Research and Development
International Journal of Engineering Research and DevelopmentInternational Journal of Engineering Research and Development
International Journal of Engineering Research and Development
 
B02120307013
B02120307013B02120307013
B02120307013
 
A trust management system for ad hoc mobile
A trust management system for ad hoc mobileA trust management system for ad hoc mobile
A trust management system for ad hoc mobile
 
Hybrid Based Resource Provisioning in Cloud
Hybrid Based Resource Provisioning in CloudHybrid Based Resource Provisioning in Cloud
Hybrid Based Resource Provisioning in Cloud
 
Extending Grids with Cloud Resource Management for Scientific Computing
Extending Grids with Cloud Resource Management for Scientific ComputingExtending Grids with Cloud Resource Management for Scientific Computing
Extending Grids with Cloud Resource Management for Scientific Computing
 
International Refereed Journal of Engineering and Science (IRJES)
International Refereed Journal of Engineering and Science (IRJES)International Refereed Journal of Engineering and Science (IRJES)
International Refereed Journal of Engineering and Science (IRJES)
 
A REVIEW ON LOAD BALANCING IN CLOUD USING ENHANCED GENETIC ALGORITHM
A REVIEW ON LOAD BALANCING IN CLOUD USING ENHANCED GENETIC ALGORITHM A REVIEW ON LOAD BALANCING IN CLOUD USING ENHANCED GENETIC ALGORITHM
A REVIEW ON LOAD BALANCING IN CLOUD USING ENHANCED GENETIC ALGORITHM
 

En vedette

Example data specifications and info requirements framework OVERVIEW
Example data specifications and info requirements framework OVERVIEWExample data specifications and info requirements framework OVERVIEW
Example data specifications and info requirements framework OVERVIEWAlan D. Duncan
 
Hr analytics – demystified!
Hr analytics – demystified!Hr analytics – demystified!
Hr analytics – demystified!Arun Krishnan
 
Stress management in hr
Stress management in hrStress management in hr
Stress management in hr'Anuraag Ghosh
 
HR Dashboard Metrics 2013
HR Dashboard Metrics 2013HR Dashboard Metrics 2013
HR Dashboard Metrics 2013nutmegslim
 
KPI for HR Manager - Sample of KPIs for HR
KPI for HR Manager - Sample of KPIs for HRKPI for HR Manager - Sample of KPIs for HR
KPI for HR Manager - Sample of KPIs for HRYodhia Antariksa
 

En vedette (8)

Example data specifications and info requirements framework OVERVIEW
Example data specifications and info requirements framework OVERVIEWExample data specifications and info requirements framework OVERVIEW
Example data specifications and info requirements framework OVERVIEW
 
HR Analytics & HR Tools
HR Analytics & HR ToolsHR Analytics & HR Tools
HR Analytics & HR Tools
 
Hr analytics – demystified!
Hr analytics – demystified!Hr analytics – demystified!
Hr analytics – demystified!
 
Hr analytics
Hr analyticsHr analytics
Hr analytics
 
HR Analytics, Done Right
HR Analytics, Done RightHR Analytics, Done Right
HR Analytics, Done Right
 
Stress management in hr
Stress management in hrStress management in hr
Stress management in hr
 
HR Dashboard Metrics 2013
HR Dashboard Metrics 2013HR Dashboard Metrics 2013
HR Dashboard Metrics 2013
 
KPI for HR Manager - Sample of KPIs for HR
KPI for HR Manager - Sample of KPIs for HRKPI for HR Manager - Sample of KPIs for HR
KPI for HR Manager - Sample of KPIs for HR
 

Similaire à Data dayposter v1.2

THE EFFECT OF THE RESOURCE CONSUMPTION CHARACTERISTICS OF CLOUD APPLICATIONS ...
THE EFFECT OF THE RESOURCE CONSUMPTION CHARACTERISTICS OF CLOUD APPLICATIONS ...THE EFFECT OF THE RESOURCE CONSUMPTION CHARACTERISTICS OF CLOUD APPLICATIONS ...
THE EFFECT OF THE RESOURCE CONSUMPTION CHARACTERISTICS OF CLOUD APPLICATIONS ...ijccsa
 
DYNAMIC ALLOCATION METHOD FOR EFFICIENT LOAD BALANCING IN VIRTUAL MACHINES FO...
DYNAMIC ALLOCATION METHOD FOR EFFICIENT LOAD BALANCING IN VIRTUAL MACHINES FO...DYNAMIC ALLOCATION METHOD FOR EFFICIENT LOAD BALANCING IN VIRTUAL MACHINES FO...
DYNAMIC ALLOCATION METHOD FOR EFFICIENT LOAD BALANCING IN VIRTUAL MACHINES FO...acijjournal
 
Optimum Resource Allocation using Specification Matching and Priority Based M...
Optimum Resource Allocation using Specification Matching and Priority Based M...Optimum Resource Allocation using Specification Matching and Priority Based M...
Optimum Resource Allocation using Specification Matching and Priority Based M...rahulmonikasharma
 
Automatic scaling of web applications for cloud computing services
Automatic scaling of web applications for cloud computing servicesAutomatic scaling of web applications for cloud computing services
Automatic scaling of web applications for cloud computing serviceseSAT Journals
 
A survey on various resource allocation policies in cloud computing environment
A survey on various resource allocation policies in cloud computing environmentA survey on various resource allocation policies in cloud computing environment
A survey on various resource allocation policies in cloud computing environmenteSAT Journals
 
Prediction Based Efficient Resource Provisioning and Its Impact on QoS Parame...
Prediction Based Efficient Resource Provisioning and Its Impact on QoS Parame...Prediction Based Efficient Resource Provisioning and Its Impact on QoS Parame...
Prediction Based Efficient Resource Provisioning and Its Impact on QoS Parame...IJECEIAES
 
Profit Maximization for Service Providers using Hybrid Pricing in Cloud Compu...
Profit Maximization for Service Providers using Hybrid Pricing in Cloud Compu...Profit Maximization for Service Providers using Hybrid Pricing in Cloud Compu...
Profit Maximization for Service Providers using Hybrid Pricing in Cloud Compu...Editor IJCATR
 
IEEE 2015-2016 A Profit Maximization Scheme with Guaranteed Quality of Servic...
IEEE 2015-2016 A Profit Maximization Scheme with Guaranteed Quality of Servic...IEEE 2015-2016 A Profit Maximization Scheme with Guaranteed Quality of Servic...
IEEE 2015-2016 A Profit Maximization Scheme with Guaranteed Quality of Servic...1crore projects
 
Using Kafka on Event-driven Microservices Architectures - Apache Kafka Meetup
Using Kafka on Event-driven Microservices Architectures - Apache Kafka MeetupUsing Kafka on Event-driven Microservices Architectures - Apache Kafka Meetup
Using Kafka on Event-driven Microservices Architectures - Apache Kafka MeetupStratio
 
IRJET- Optimization of Completion Time through Efficient Resource Allocation ...
IRJET- Optimization of Completion Time through Efficient Resource Allocation ...IRJET- Optimization of Completion Time through Efficient Resource Allocation ...
IRJET- Optimization of Completion Time through Efficient Resource Allocation ...IRJET Journal
 
PROPOSED ONTOLOGY FRAMEWORK FOR DYNAMIC RESOURCE PROVISIONING ON PUBLIC CLOUD
PROPOSED ONTOLOGY FRAMEWORK FOR DYNAMIC RESOURCE PROVISIONING ON PUBLIC CLOUDPROPOSED ONTOLOGY FRAMEWORK FOR DYNAMIC RESOURCE PROVISIONING ON PUBLIC CLOUD
PROPOSED ONTOLOGY FRAMEWORK FOR DYNAMIC RESOURCE PROVISIONING ON PUBLIC CLOUDIAEME Publication
 
Proactive Scheduling in Cloud Computing
Proactive Scheduling in Cloud ComputingProactive Scheduling in Cloud Computing
Proactive Scheduling in Cloud ComputingjournalBEEI
 
construction management.pptx
construction management.pptxconstruction management.pptx
construction management.pptxpraful91
 
Intelligent Workload Management in Virtualized Cloud Environment
Intelligent Workload Management in Virtualized Cloud EnvironmentIntelligent Workload Management in Virtualized Cloud Environment
Intelligent Workload Management in Virtualized Cloud EnvironmentIJTET Journal
 
IMPACT OF RESOURCE MANAGEMENT AND SCALABILITY ON PERFORMANCE OF CLOUD APPLICA...
IMPACT OF RESOURCE MANAGEMENT AND SCALABILITY ON PERFORMANCE OF CLOUD APPLICA...IMPACT OF RESOURCE MANAGEMENT AND SCALABILITY ON PERFORMANCE OF CLOUD APPLICA...
IMPACT OF RESOURCE MANAGEMENT AND SCALABILITY ON PERFORMANCE OF CLOUD APPLICA...IJCSEA Journal
 
IMPACT OF RESOURCE MANAGEMENT AND SCALABILITY ON PERFORMANCE OF CLOUD APPLICA...
IMPACT OF RESOURCE MANAGEMENT AND SCALABILITY ON PERFORMANCE OF CLOUD APPLICA...IMPACT OF RESOURCE MANAGEMENT AND SCALABILITY ON PERFORMANCE OF CLOUD APPLICA...
IMPACT OF RESOURCE MANAGEMENT AND SCALABILITY ON PERFORMANCE OF CLOUD APPLICA...IJCSEA Journal
 
IMPACT OF RESOURCE MANAGEMENT AND SCALABILITY ON PERFORMANCE OF CLOUD APPLICA...
IMPACT OF RESOURCE MANAGEMENT AND SCALABILITY ON PERFORMANCE OF CLOUD APPLICA...IMPACT OF RESOURCE MANAGEMENT AND SCALABILITY ON PERFORMANCE OF CLOUD APPLICA...
IMPACT OF RESOURCE MANAGEMENT AND SCALABILITY ON PERFORMANCE OF CLOUD APPLICA...IJCSEA Journal
 
IMPACT OF RESOURCE MANAGEMENT AND SCALABILITY ON PERFORMANCE OF CLOUD APPLICA...
IMPACT OF RESOURCE MANAGEMENT AND SCALABILITY ON PERFORMANCE OF CLOUD APPLICA...IMPACT OF RESOURCE MANAGEMENT AND SCALABILITY ON PERFORMANCE OF CLOUD APPLICA...
IMPACT OF RESOURCE MANAGEMENT AND SCALABILITY ON PERFORMANCE OF CLOUD APPLICA...IJCSEA Journal
 

Similaire à Data dayposter v1.2 (20)

THE EFFECT OF THE RESOURCE CONSUMPTION CHARACTERISTICS OF CLOUD APPLICATIONS ...
THE EFFECT OF THE RESOURCE CONSUMPTION CHARACTERISTICS OF CLOUD APPLICATIONS ...THE EFFECT OF THE RESOURCE CONSUMPTION CHARACTERISTICS OF CLOUD APPLICATIONS ...
THE EFFECT OF THE RESOURCE CONSUMPTION CHARACTERISTICS OF CLOUD APPLICATIONS ...
 
DYNAMIC ALLOCATION METHOD FOR EFFICIENT LOAD BALANCING IN VIRTUAL MACHINES FO...
DYNAMIC ALLOCATION METHOD FOR EFFICIENT LOAD BALANCING IN VIRTUAL MACHINES FO...DYNAMIC ALLOCATION METHOD FOR EFFICIENT LOAD BALANCING IN VIRTUAL MACHINES FO...
DYNAMIC ALLOCATION METHOD FOR EFFICIENT LOAD BALANCING IN VIRTUAL MACHINES FO...
 
Optimum Resource Allocation using Specification Matching and Priority Based M...
Optimum Resource Allocation using Specification Matching and Priority Based M...Optimum Resource Allocation using Specification Matching and Priority Based M...
Optimum Resource Allocation using Specification Matching and Priority Based M...
 
Automatic scaling of web applications for cloud computing services
Automatic scaling of web applications for cloud computing servicesAutomatic scaling of web applications for cloud computing services
Automatic scaling of web applications for cloud computing services
 
A survey on various resource allocation policies in cloud computing environment
A survey on various resource allocation policies in cloud computing environmentA survey on various resource allocation policies in cloud computing environment
A survey on various resource allocation policies in cloud computing environment
 
Prediction Based Efficient Resource Provisioning and Its Impact on QoS Parame...
Prediction Based Efficient Resource Provisioning and Its Impact on QoS Parame...Prediction Based Efficient Resource Provisioning and Its Impact on QoS Parame...
Prediction Based Efficient Resource Provisioning and Its Impact on QoS Parame...
 
Profit Maximization for Service Providers using Hybrid Pricing in Cloud Compu...
Profit Maximization for Service Providers using Hybrid Pricing in Cloud Compu...Profit Maximization for Service Providers using Hybrid Pricing in Cloud Compu...
Profit Maximization for Service Providers using Hybrid Pricing in Cloud Compu...
 
Presentation
PresentationPresentation
Presentation
 
IEEE 2015-2016 A Profit Maximization Scheme with Guaranteed Quality of Servic...
IEEE 2015-2016 A Profit Maximization Scheme with Guaranteed Quality of Servic...IEEE 2015-2016 A Profit Maximization Scheme with Guaranteed Quality of Servic...
IEEE 2015-2016 A Profit Maximization Scheme with Guaranteed Quality of Servic...
 
Using Kafka on Event-driven Microservices Architectures - Apache Kafka Meetup
Using Kafka on Event-driven Microservices Architectures - Apache Kafka MeetupUsing Kafka on Event-driven Microservices Architectures - Apache Kafka Meetup
Using Kafka on Event-driven Microservices Architectures - Apache Kafka Meetup
 
IRJET- Optimization of Completion Time through Efficient Resource Allocation ...
IRJET- Optimization of Completion Time through Efficient Resource Allocation ...IRJET- Optimization of Completion Time through Efficient Resource Allocation ...
IRJET- Optimization of Completion Time through Efficient Resource Allocation ...
 
PROPOSED ONTOLOGY FRAMEWORK FOR DYNAMIC RESOURCE PROVISIONING ON PUBLIC CLOUD
PROPOSED ONTOLOGY FRAMEWORK FOR DYNAMIC RESOURCE PROVISIONING ON PUBLIC CLOUDPROPOSED ONTOLOGY FRAMEWORK FOR DYNAMIC RESOURCE PROVISIONING ON PUBLIC CLOUD
PROPOSED ONTOLOGY FRAMEWORK FOR DYNAMIC RESOURCE PROVISIONING ON PUBLIC CLOUD
 
Scheduling in CCE
Scheduling in CCEScheduling in CCE
Scheduling in CCE
 
Proactive Scheduling in Cloud Computing
Proactive Scheduling in Cloud ComputingProactive Scheduling in Cloud Computing
Proactive Scheduling in Cloud Computing
 
construction management.pptx
construction management.pptxconstruction management.pptx
construction management.pptx
 
Intelligent Workload Management in Virtualized Cloud Environment
Intelligent Workload Management in Virtualized Cloud EnvironmentIntelligent Workload Management in Virtualized Cloud Environment
Intelligent Workload Management in Virtualized Cloud Environment
 
IMPACT OF RESOURCE MANAGEMENT AND SCALABILITY ON PERFORMANCE OF CLOUD APPLICA...
IMPACT OF RESOURCE MANAGEMENT AND SCALABILITY ON PERFORMANCE OF CLOUD APPLICA...IMPACT OF RESOURCE MANAGEMENT AND SCALABILITY ON PERFORMANCE OF CLOUD APPLICA...
IMPACT OF RESOURCE MANAGEMENT AND SCALABILITY ON PERFORMANCE OF CLOUD APPLICA...
 
IMPACT OF RESOURCE MANAGEMENT AND SCALABILITY ON PERFORMANCE OF CLOUD APPLICA...
IMPACT OF RESOURCE MANAGEMENT AND SCALABILITY ON PERFORMANCE OF CLOUD APPLICA...IMPACT OF RESOURCE MANAGEMENT AND SCALABILITY ON PERFORMANCE OF CLOUD APPLICA...
IMPACT OF RESOURCE MANAGEMENT AND SCALABILITY ON PERFORMANCE OF CLOUD APPLICA...
 
IMPACT OF RESOURCE MANAGEMENT AND SCALABILITY ON PERFORMANCE OF CLOUD APPLICA...
IMPACT OF RESOURCE MANAGEMENT AND SCALABILITY ON PERFORMANCE OF CLOUD APPLICA...IMPACT OF RESOURCE MANAGEMENT AND SCALABILITY ON PERFORMANCE OF CLOUD APPLICA...
IMPACT OF RESOURCE MANAGEMENT AND SCALABILITY ON PERFORMANCE OF CLOUD APPLICA...
 
IMPACT OF RESOURCE MANAGEMENT AND SCALABILITY ON PERFORMANCE OF CLOUD APPLICA...
IMPACT OF RESOURCE MANAGEMENT AND SCALABILITY ON PERFORMANCE OF CLOUD APPLICA...IMPACT OF RESOURCE MANAGEMENT AND SCALABILITY ON PERFORMANCE OF CLOUD APPLICA...
IMPACT OF RESOURCE MANAGEMENT AND SCALABILITY ON PERFORMANCE OF CLOUD APPLICA...
 

Dernier

Online banking management system project.pdf
Online banking management system project.pdfOnline banking management system project.pdf
Online banking management system project.pdfKamal Acharya
 
Java Programming :Event Handling(Types of Events)
Java Programming :Event Handling(Types of Events)Java Programming :Event Handling(Types of Events)
Java Programming :Event Handling(Types of Events)simmis5
 
FULL ENJOY Call Girls In Mahipalpur Delhi Contact Us 8377877756
FULL ENJOY Call Girls In Mahipalpur Delhi Contact Us 8377877756FULL ENJOY Call Girls In Mahipalpur Delhi Contact Us 8377877756
FULL ENJOY Call Girls In Mahipalpur Delhi Contact Us 8377877756dollysharma2066
 
Call Girls Walvekar Nagar Call Me 7737669865 Budget Friendly No Advance Booking
Call Girls Walvekar Nagar Call Me 7737669865 Budget Friendly No Advance BookingCall Girls Walvekar Nagar Call Me 7737669865 Budget Friendly No Advance Booking
Call Girls Walvekar Nagar Call Me 7737669865 Budget Friendly No Advance Bookingroncy bisnoi
 
VIP Model Call Girls Kothrud ( Pune ) Call ON 8005736733 Starting From 5K to ...
VIP Model Call Girls Kothrud ( Pune ) Call ON 8005736733 Starting From 5K to ...VIP Model Call Girls Kothrud ( Pune ) Call ON 8005736733 Starting From 5K to ...
VIP Model Call Girls Kothrud ( Pune ) Call ON 8005736733 Starting From 5K to ...SUHANI PANDEY
 
Double rodded leveling 1 pdf activity 01
Double rodded leveling 1 pdf activity 01Double rodded leveling 1 pdf activity 01
Double rodded leveling 1 pdf activity 01KreezheaRecto
 
Call for Papers - Educational Administration: Theory and Practice, E-ISSN: 21...
Call for Papers - Educational Administration: Theory and Practice, E-ISSN: 21...Call for Papers - Educational Administration: Theory and Practice, E-ISSN: 21...
Call for Papers - Educational Administration: Theory and Practice, E-ISSN: 21...Christo Ananth
 
Extrusion Processes and Their Limitations
Extrusion Processes and Their LimitationsExtrusion Processes and Their Limitations
Extrusion Processes and Their Limitations120cr0395
 
Top Rated Pune Call Girls Budhwar Peth ⟟ 6297143586 ⟟ Call Me For Genuine Se...
Top Rated  Pune Call Girls Budhwar Peth ⟟ 6297143586 ⟟ Call Me For Genuine Se...Top Rated  Pune Call Girls Budhwar Peth ⟟ 6297143586 ⟟ Call Me For Genuine Se...
Top Rated Pune Call Girls Budhwar Peth ⟟ 6297143586 ⟟ Call Me For Genuine Se...Call Girls in Nagpur High Profile
 
KubeKraft presentation @CloudNativeHooghly
KubeKraft presentation @CloudNativeHooghlyKubeKraft presentation @CloudNativeHooghly
KubeKraft presentation @CloudNativeHooghlysanyuktamishra911
 
ONLINE FOOD ORDER SYSTEM PROJECT REPORT.pdf
ONLINE FOOD ORDER SYSTEM PROJECT REPORT.pdfONLINE FOOD ORDER SYSTEM PROJECT REPORT.pdf
ONLINE FOOD ORDER SYSTEM PROJECT REPORT.pdfKamal Acharya
 
Unit 1 - Soil Classification and Compaction.pdf
Unit 1 - Soil Classification and Compaction.pdfUnit 1 - Soil Classification and Compaction.pdf
Unit 1 - Soil Classification and Compaction.pdfRagavanV2
 
VIP Call Girls Ankleshwar 7001035870 Whatsapp Number, 24/07 Booking
VIP Call Girls Ankleshwar 7001035870 Whatsapp Number, 24/07 BookingVIP Call Girls Ankleshwar 7001035870 Whatsapp Number, 24/07 Booking
VIP Call Girls Ankleshwar 7001035870 Whatsapp Number, 24/07 Bookingdharasingh5698
 
UNIT-II FMM-Flow Through Circular Conduits
UNIT-II FMM-Flow Through Circular ConduitsUNIT-II FMM-Flow Through Circular Conduits
UNIT-II FMM-Flow Through Circular Conduitsrknatarajan
 
data_management_and _data_science_cheat_sheet.pdf
data_management_and _data_science_cheat_sheet.pdfdata_management_and _data_science_cheat_sheet.pdf
data_management_and _data_science_cheat_sheet.pdfJiananWang21
 
Intze Overhead Water Tank Design by Working Stress - IS Method.pdf
Intze Overhead Water Tank  Design by Working Stress - IS Method.pdfIntze Overhead Water Tank  Design by Working Stress - IS Method.pdf
Intze Overhead Water Tank Design by Working Stress - IS Method.pdfSuman Jyoti
 
Vivazz, Mieres Social Housing Design Spain
Vivazz, Mieres Social Housing Design SpainVivazz, Mieres Social Housing Design Spain
Vivazz, Mieres Social Housing Design Spaintimesproduction05
 

Dernier (20)

Online banking management system project.pdf
Online banking management system project.pdfOnline banking management system project.pdf
Online banking management system project.pdf
 
Roadmap to Membership of RICS - Pathways and Routes
Roadmap to Membership of RICS - Pathways and RoutesRoadmap to Membership of RICS - Pathways and Routes
Roadmap to Membership of RICS - Pathways and Routes
 
(INDIRA) Call Girl Bhosari Call Now 8617697112 Bhosari Escorts 24x7
(INDIRA) Call Girl Bhosari Call Now 8617697112 Bhosari Escorts 24x7(INDIRA) Call Girl Bhosari Call Now 8617697112 Bhosari Escorts 24x7
(INDIRA) Call Girl Bhosari Call Now 8617697112 Bhosari Escorts 24x7
 
Java Programming :Event Handling(Types of Events)
Java Programming :Event Handling(Types of Events)Java Programming :Event Handling(Types of Events)
Java Programming :Event Handling(Types of Events)
 
FULL ENJOY Call Girls In Mahipalpur Delhi Contact Us 8377877756
FULL ENJOY Call Girls In Mahipalpur Delhi Contact Us 8377877756FULL ENJOY Call Girls In Mahipalpur Delhi Contact Us 8377877756
FULL ENJOY Call Girls In Mahipalpur Delhi Contact Us 8377877756
 
Call Girls Walvekar Nagar Call Me 7737669865 Budget Friendly No Advance Booking
Call Girls Walvekar Nagar Call Me 7737669865 Budget Friendly No Advance BookingCall Girls Walvekar Nagar Call Me 7737669865 Budget Friendly No Advance Booking
Call Girls Walvekar Nagar Call Me 7737669865 Budget Friendly No Advance Booking
 
VIP Model Call Girls Kothrud ( Pune ) Call ON 8005736733 Starting From 5K to ...
VIP Model Call Girls Kothrud ( Pune ) Call ON 8005736733 Starting From 5K to ...VIP Model Call Girls Kothrud ( Pune ) Call ON 8005736733 Starting From 5K to ...
VIP Model Call Girls Kothrud ( Pune ) Call ON 8005736733 Starting From 5K to ...
 
Double rodded leveling 1 pdf activity 01
Double rodded leveling 1 pdf activity 01Double rodded leveling 1 pdf activity 01
Double rodded leveling 1 pdf activity 01
 
Call for Papers - Educational Administration: Theory and Practice, E-ISSN: 21...
Call for Papers - Educational Administration: Theory and Practice, E-ISSN: 21...Call for Papers - Educational Administration: Theory and Practice, E-ISSN: 21...
Call for Papers - Educational Administration: Theory and Practice, E-ISSN: 21...
 
Extrusion Processes and Their Limitations
Extrusion Processes and Their LimitationsExtrusion Processes and Their Limitations
Extrusion Processes and Their Limitations
 
Top Rated Pune Call Girls Budhwar Peth ⟟ 6297143586 ⟟ Call Me For Genuine Se...
Top Rated  Pune Call Girls Budhwar Peth ⟟ 6297143586 ⟟ Call Me For Genuine Se...Top Rated  Pune Call Girls Budhwar Peth ⟟ 6297143586 ⟟ Call Me For Genuine Se...
Top Rated Pune Call Girls Budhwar Peth ⟟ 6297143586 ⟟ Call Me For Genuine Se...
 
KubeKraft presentation @CloudNativeHooghly
KubeKraft presentation @CloudNativeHooghlyKubeKraft presentation @CloudNativeHooghly
KubeKraft presentation @CloudNativeHooghly
 
ONLINE FOOD ORDER SYSTEM PROJECT REPORT.pdf
ONLINE FOOD ORDER SYSTEM PROJECT REPORT.pdfONLINE FOOD ORDER SYSTEM PROJECT REPORT.pdf
ONLINE FOOD ORDER SYSTEM PROJECT REPORT.pdf
 
Unit 1 - Soil Classification and Compaction.pdf
Unit 1 - Soil Classification and Compaction.pdfUnit 1 - Soil Classification and Compaction.pdf
Unit 1 - Soil Classification and Compaction.pdf
 
VIP Call Girls Ankleshwar 7001035870 Whatsapp Number, 24/07 Booking
VIP Call Girls Ankleshwar 7001035870 Whatsapp Number, 24/07 BookingVIP Call Girls Ankleshwar 7001035870 Whatsapp Number, 24/07 Booking
VIP Call Girls Ankleshwar 7001035870 Whatsapp Number, 24/07 Booking
 
UNIT-II FMM-Flow Through Circular Conduits
UNIT-II FMM-Flow Through Circular ConduitsUNIT-II FMM-Flow Through Circular Conduits
UNIT-II FMM-Flow Through Circular Conduits
 
data_management_and _data_science_cheat_sheet.pdf
data_management_and _data_science_cheat_sheet.pdfdata_management_and _data_science_cheat_sheet.pdf
data_management_and _data_science_cheat_sheet.pdf
 
NFPA 5000 2024 standard .
NFPA 5000 2024 standard                                  .NFPA 5000 2024 standard                                  .
NFPA 5000 2024 standard .
 
Intze Overhead Water Tank Design by Working Stress - IS Method.pdf
Intze Overhead Water Tank  Design by Working Stress - IS Method.pdfIntze Overhead Water Tank  Design by Working Stress - IS Method.pdf
Intze Overhead Water Tank Design by Working Stress - IS Method.pdf
 
Vivazz, Mieres Social Housing Design Spain
Vivazz, Mieres Social Housing Design SpainVivazz, Mieres Social Housing Design Spain
Vivazz, Mieres Social Housing Design Spain
 

Data dayposter v1.2

  • 1. TEMPLATE DESIGN © 2008 www.PosterPresentations.com Auto-Scaling Frameworks for Big Data Analytics on Clouds  The primary goal of big data analytics is to help companies make more informed business decisions.  Uncover hidden patterns and unknown correlations between data.  High-performance analytics necessary to determine the relevance of data.  Gartner states that through 2015, 85% of Fortune 500 organizations will be unable to exploit big data for competitive advantage. [1]  Traditional data processing tools cannot handle a large volume of data.  Google’s MapReduce programming model is able to handle big data.  MapReduce requires multiple compute resources to process large tasks.  The demand to process data varies based on the current period.  A MapReduce model can be used together with Cloud Computing.  Cloud computing enables companies to consume compute resources as a utility rather building and maintaining computing infrastructures in-house.  To handle variable load while processing data, a technique known as Auto-Scaling is employed.  An Intermediary enterprise handles user jobs from a single client enterprise and executes those jobs on resources acquired from a public cloud provider to form a virtual private cloud for the users.  The broker is hosted in the intermediary enterprise.  The public cloud provider charges the broker c_pub dollars an hour per resource.  The broker charges the user c_pvt dollars per second.  A novel proactive [3] and a reactive [4] auto-scaling framework which uses a price model that can lead to an increase in the profit for the broker (intermediary enterprise) and a reduction in the user cost at the same time.  Both frameworks handle SLA-driven advanced reservation requests as well as well as On-Demand requests.  A detailed performance analysis focusing on broker profit and user cost for a prototype subjected to various combinations of system and workload parameters is performed and key insights into system behavior is presented.  System II: A system that can scale up if it can not fit a request on the resources available.  Resources are scaled down when they are no longer needed.  Comparing broker profit and total user cost of System II with System I.  System III: The users purchase resources directly from Amazon.  There is no broker present in this system.  Comparing total user cost of System III with System I and System II.  Auto-Scaling is the ability to modify the capacity for the user’s cloud infrastructure based on traffic patterns.  Benefits of Auto-Scaling:  Fault tolerant  Highly available  No Capacity Planning  Amazon’s CloudWatch allow users to set thresholds for Auto-Scaling.  Qubole [2] offer cloud based data analytics services based on the MapReduce programming Model with Apache’s Hadoop and Hive.  Qubole uses resources from the Amazon Cloud and allows auto-scaling based on user’s workload.  Qubole offers cost savings by optimizing data processing tasks.  However, these services do not handle deadlines with individual user jobs which perform the data analytics tasks. Two types of Auto-Scaling techniques are discussed:  A proactive framework that scales resources based on predictions from a machine learning engine (MLE). This system scales independent off the user demand.  A reactive framework that scales resources when a request arrives to the system. This system scales depending on the user demand. Effect of Load Factor on BP and UC  Load Factor (f) – is the ratio of the number of requests generated with an arrival rate of λlow to the total number of requests generated during the experiment. Performance metrics:  Broker profit (BP) - is the profit a broker earns ($/hour )  Total User cost (UC) - is the amount charged to the user ($/hour)  The number of resources in the pool used by the user requests need not be determined a priori and are controlled dynamically thereby reducing the cost for capacity planning.  Overall, System I preforms better than System II and System III in terms of Broker Profit and User Cost.  The Auto-Scaling frameworks accept a MapReduce workload that contain jobs to analyze big data. Other types of workloads may also be handled by the Auto-Scaling system.  Both frameworks allow users in an enterprise for example, to submit Advance Reservation as well as On-Demand requests to a private cloud provided by the intermediary cloud provider.  The reactive framework provides a minimum grade of service to ensure that a certain percentage of user requests are guaranteed to be executed. Introduction 1 Contributions 5 The Broker Model 3 Other Systems for Performance Evaluations 7 Conclusions 9 This research is supported by the Natural Sciences and Engineering Council of Canada (NSERC) and TELUS ---------------------------------------------------------------------------------------------------- [1] Gartner Report, Big Data Management & Analytics, http://www.gartner.com/technology/topics/big-data.jsp. [2] Qubole, http://www.qubole.com/, Accessed March 2015. [3] A. Biswas, S. Majumdar, B. Nandy and A. El-Haraki, "Automatic Resource Provisioning: a Machine Learning based Proactive approach," in Proc. of International Conference on Cloud Computing Technology and Science (CloudCom), Singapore, 12 2014. [4] A. Biswas, S. Majumdar, B. Nandy and A. El-Haraki, "An Auto-scaling Framework for Controlling Enterprise Resources on Clouds," in Proc. of the 15th IEEE/ACM International Symposium on Cluster, Cloud and Grid Computing (CCGrid), Shenzhen, 05, 2015. Acknowledgements and References 11 Auto-Scaling 2 Auto-scaling decisions can be made: Reactively: the system reacts to changes in user workload.  Scaling conditions based on a target metric reaching some threshold.  Set a threshold for a metric which triggers an adjustment of resources. Proactively: the system tries to predict future resource requirements in order to ensure sufficient resource is available ahead of time.  Based on time series analysis or control theory.  Predict the workload based on past demand. Types of Auto-Scaling 3 http://docs.aws.amazon.com/AutoScaling/latest/DeveloperGuide/ WhatIsAutoScaling.html Section “What is Auto Scaling” 4 Internet Single Client Enterprise Broker Virtual Private Cloud Resources End Users Public Cloud Provider Intermediary Cloud Provider Auto-Scaling frameworks 6 Sample Performance Results (Proactive System) 8  BP decreases with f. System I earns more profit that System II.  UC decreases with f. System I charges users a lesser amount that System II and System III.  BP increases with increase in Service Time and Laxity Factor and System I earns more profit that System II. Effect of Service Time and Laxity on BP Anshuman Biswas Systems and Computer Engineering Carleton University anshuman@sce.carleton.ca Supervisors: Shikharesh Majumdar, Biswajit Nandy Systems and Computer Engineering Carleton University {majumdar,bnandy}@sce.carleton.ca Collaborator: Ali El-Haraki TELUS User requests have Service Level Agreements (SLAs): Arrival Time, Earliest start time , Execution time and Deadline RequestHandler User Request MatchMake Sched Reactive Autoscaler Dynamic Resouce Pool Manager Decision Maker GoS Decision Maker Predictive Autoscaler Dynamic Resource Pool Manager RequestHandler Machine Learning Engine MatchMake Sched User Request Reactive Auto-Scaler Proactive Auto-Scaler The Reactive Auto-Scaler:  Request Handler (RH) is responsible for handling user requests.  RH forwards this request to Decision Maker (DM) which decided whether to accept or reject the request based on two operations. DM sends the request to the Matchmaking and scheduling component (MMS) which is responsible for matchmaking and scheduling the request. If MMS cannot schedule the request, DM then invokes the reactive auto-scaler.  A criterion for the reactive Auto-Scaler for acquiring resources is based on grade of service (GoS). The GoS is specified by the client enterprise. The system described uses the blocking ratio (B) as the GoS. B is the proportion of requests that cannot be completed before the expiry of their deadlines and are therefore rejected by the system.  For the reactive Auto-Scaler, DM consults with a GoS.  Framework provides a minimum Quality of Service (QoS) for the users.  Acquiring and releasing a resource is based on these three rules: Rule I: When B > Bspec acquire resource, where Bspec is the desired value of B (GoS) maintained by the broker.  Rule II: When (Broker Profit)i > 0, acquire ith resource. Rule III: When (stopi) = current time, release ith resource, where stopi is the stop time for the ith resource. The Proactive Auto-Scaler:  This broker replaces the GoS module with the Machine Learning Engine (MLE) Module.  MLE predicts the requests arriving in the future and then auto-scales.  Auto-scaling resources is based on Rule II and Rule III from the Reactive broker.  MLE predicts after a set number of requests have arrived in the system. The reactive/proactive auto-scaling system is known as System I. Future Work 10 Detailed performance analysis using reactive auto-scaling technique. Additional simulation experiments for various other combination of system and workload parameters are being planned. Using traces of real workload in simulation can lead to interesting insights into system behaviour and performance.