You are on page 1of 40

Optimal Utilization based VM Consolidation

Technique for Green Cloud Computing

Department of Computer Science & Engineering


Under Guidance of
Presented By
Prof. Sapna Choudhary
Priyanka Nema
Enrol No.-0228CS13MT12
Department of CSE

HOD

SHRI RAM GROUP OF INSTITUTION,JABALPUR (M.P.)

OUTLINE
Introduction
Problem Statement
Objectives
Literature survey
Proposed Approach
Experimental Setup
Result Analysis
Conclusions & Future Scope
Publication Details
References

Introduction
What is cloud computing???

Cloud computing is a model which rapidly


delivered and released with minimal management
effort or service provider interaction. It enables
convenient, on-demand network access to a
shared collection of configurable computing
resources

ATTRIBUTES FULFILLED BY CLOUD COMPUTING

Add Figure for features here

CLOUD SERVICE DELIVERY


MODELS

SaaS( Software as a Service)


PaaS(Platform as a Service)
IaaS(Infrastructure as a Service)

Add figure

Service Delivery models

Add Introductory topics very


relevant to your work
Like VM Live Migration etc
,,,,,,,,,,,,,,,,,,,,,

Problem Statement
Change

as per your work


In cloud computing the access to the resources or
delivery is dynamic, and on demand.
This increase power consumption, increased
operational cost and high carbon dioxide emission.
The challenge is by satisfying high QoS (Quality of
service) and SLA (Service level agreements)
requirements with maximizing profits.
Dynamic virtual machine consolidation is potential
approach for reducing energy consumption.

Objectives
Change

as per your work


To propose the SLA and energy aware dynamic VM
consolidation policy and provide the baseline for better
performance and environment.
Includes

determination of overload and underload


hosts, so that effective decision like migration of VMs to
other hosts, can be taken to optimize power.
By

introducing a proposed algorithm, a comparative


study is done with the proposed and various existing
energy VM techniques.

SLA And Energy Consumption


Stands for Service level agreement
The SLA is a contract between consumer and
provider that describes the services agreed like
service level parameters, guarantees and actions
and consequences for all case of violations.

Consolidation Policies
Migration policies

Literature survey

Add as per our references


Anton Beloglazov et al.presented a survey of
research in energy-efficient computing. The
architectural
principles
for
energy-efficient
management of Clouds; energy-efficient resource
allocation policies and scheduling algorithms
considering QoS expectations and power usage
characteristics of the devices; and a number of open
research challenges, are addressed. This works
substantially contribute to both resource providers
and consumers

Literature survey
Add as per our references
Beloglazov and Buyya have proposed a novel
technique for dynamic consolidation of VMs based
on automatically adjusting (adaptive) utilization
thresholds, which ensures a high level of meeting
the Service Level Agreements (SLA). The SLA
violation is less than 1% and good results achieved
in number of VM migrations and energy
consumption.

Literature survey
Add as per our references
Yuxiang Shi et al.have achieved the energy
efficiency objective by dynamically allocating
resources based on utilization analysis from
resource utilization log and prediction methods
known as Linear Predicting Method (LPM) and
Flat Period Reservation-Reduced Method
(FPRRM). The M/M/1 queuing theory is used for
predicting method and has better response time
and less energy-consuming.

Outline of Proposed Approach


System Model
Proposed Framework
Experimental setup
Parameters
General Algorithm
Proposed Algorithm
VM .. policy
Formula Used
Parameters Used
Existing .. Policies
Output Snapshot
Result Analysis

Proposed Approach

Add figure for System MOdel /

System Model

Proposed Framework

Add figure of Proposed Framework with small


explanation

Experimental Setup

Add if necessary and as time availability for ppt


About Tool (CloudSim 3.0.2)
Functionalities: It supports modeling and simulation of:
Cloud computing data centers and data center network topologies at
large scale.
customizable policies for provisioning of host resources to virtual
machines on virtualized server hosts.
energy-aware computational resources.
message-passing applications and federated clouds.
elements dynamically included with stop and resume facility in
simulation.
user-defined allocation policies of hosts to virtual machines and policies
for allocation of host resources to virtual machines.

General Simulaton Parameters


Variables
Planet Lab workload
NUMBER_OF_HOSTS
No. of VMs
VM Allocation Policy
VM Selection Policy
Ram utilization threshold
CPU utilization threshold
SCHEDULING_INTERVAL
SIMULATION_LIMIT

Value
Add yourself so that
you remember

Remark

Configuration parameters
Datacenter

Architecture

X86

Operating System

Linux

Virtual Machine Manager

Xen

Types

02

Type 1:
[Xeon 3040 1860 MHz, 2 cores], 4GB);
Type 2:
[Xeon 3075 2660 MHz, 2 cores], 4GB);
Host

MIPS

1860, 2660 (MHz)

PE

2, 2

RAM

4, 4 (GB)

Storage

1 GB

Bandwidth

1 Gbit/s

VM Scheduler

Time shared

Configuration parameters

Virtual Machine

Cloudlet / Task

Types

04

MIPS

2500, 2000, 1000, 500

Processing Elements

1, 1, 1, 1

RAM

870, 1740, 1740, 613 (MB)

Size

2.5 GB

Bandwidth

100 Mbit/s

Cloud Let Scheduler

Time Shared

Virtual Machine Manager

Xen

Length

2500 * SIMULATION_LIMIT

Full Size

300 MB

Output Size

300 Mb

Processing Elements

General Algorithm

Add if necessary as per work


Input: Host List (containing m hosts)
For each host (1 to m)
if (host is Overloaded) then
Select and Add VMs to migrate to VM List from this Overloaded
host;
Place VMs to Host from VM List;
End if
For each host (1 to m)
if (host is Underloaded) then
Select and Add VMs to migrate, to VM List from this
Underloaded host;
Place VMs to Host from VM List;
End if
End for
End for

Proposed Algorithm
For

VM Selection / Migration Policy (Name of proposed)


The

next step after the determination of


overloaded host is determination of the particular
VMs to migrate from that host.
We propose a policy for VM selection. Our
framework outperformed by the use of proposed
Proposed Policy for VM .
The
existing
policy
only
considers
the...
The
proposed
policy
is
considers...

Formula Used
v= Virtual Machine
Vj= Set of VMs Currently Allocated to host j
RAMu(v)= Amount of RAM migrated by VM v
RAMu(a)= Amount of RAM utilized by the VM a
NETj = Spare network bandwidth available for host j

is total requested RAM, by n virtual machines.

PERFORMANCE METRICS
Total energy consumption:
Total energy consumption is defined as the sum of
energy consumed by the physical resources of a
data center as a result of application workloads.
SLA violation:
When a VM cannot get the promised Quality of
Service (QoS), SLA violation takes place. For
example when a VM cannot get requested MIPS
SLA violation issue occur.

Number of VM migrations:
For dynamic VM consolidation it is must to
determine overloaded and under-loaded hosts and
once the overloaded or under-loaded hosts found the
VMs get selected for migration. The minimization of
the VM migration time is more important constraint
and one of the ways to achieve it is to reduce the total
number of VM
migrations.

Consolidation Policies
CONSOLIDATION POLICY

CONVENTION

Minimum Utilization+ Threshold Algorithm

ThrMu

Minimum Utilization + Median Absolute Deviation

MadMu

Minimum Utilization + Interquartile Range

IqrMu

Minimum Migration Time + Threshold Algorithm

ThrMmt

Minimum Migration Time + Median Absolute Deviation

MadMmt

Minimum Migration Time + Interquartile Range

IqrMmt

Proposed

Name

Output Screenshot
ThrMmt Output

Add obtained
screenshots..
Proposed Algo Output

Result Analysis
Parameters

Dvfs

Energy Consumption (kWh)

Add obtained Graph


here..

IqrMmt

MadMmt

ThrMmt

Proposed

Result Analysis
Parameters

Dvfs

Average SLA violation (%)

Add obtained Graph


here..

IqrMmt

MadMmt

ThrMmt

Proposed

Suggestions for ppt

Use less theory and more pictures. Justify text in slides. Use white / cream slide and bright
and colourful fonts. Use Times New Roman or such fonts.. Use good bullets.
Avoid using jumbled paragraph (too much theory), instead use points and figures. E.g. use
figures for Public, private, hybrid cloud and IaaS, PaaS, SaaS.
Explain consolidation etc. in short before problem statement.
Use references in literature survey.
Explain system model with figure [12, 13, 14, 7]. It would be better to draw your own figure
for report.
Now this ppt have enough slides to understand proposed algorithm/policy/method. You may
arrange it in more efficient way.
Also add screenshots of program and / or output.

Result Analysis
Parameters
No. of VM migrations

Add obtained Graph


here..

Dvfs

IqrMmt MadMm ThrMmt


t

Proposed

Conclusions

Modify as per our theoretical and Practical results


This work explored that the problem of host overload and
under-load detection is part of dynamic VM consolidation
and generally rely on statistical analysis of historical data.
This work focuses on:
Energy-efficient computing;
Presented a Survey on SLA and Energy-Efficient Dynamic Virtual
Machine (VM) Consolidation/Migration/selection;
Proposed a ..
Presented a Comparative analysis based on a performance
evaluation study conducted on various energy efficient VM
consolidation techniques using real world workload traces, on
basis of various heuristics.

Future Scope

Modify as per our theoretical and Practical results

This work suggests the study of other energy


efficient Resource provisioning techniques over
varied workloads to make the data centers scalable
and reliable in terms of QoS
It suggests the analysis and modification to
energy-efficient dynamic VM consolidation in
terms of host overload detection; VM selection etc.
individually.

Publication Details

References
P. Mell, T. Grance, The NIST Definition of Cloud
Computing, National Institute of Standards and
Technology, ver. 15, 9 July 2010.
Saurabh Kumar Garg, Steve Versteeg, Rajkumar
Buyyaa, A framework for ranking of cloud
computing services, Future Generation Computer
Systems 29 (2013) 10121023, Elsevier 2012.
Etc
Etc

QUESTIONS

THANK YOU

You might also like