Task Scheduling Using Firefly algorithm with cloudsim

Task Scheduling Using Firefly Algorithm in
Cloud Computing Environment Using Cloudsim
Name : Ahmad Aqil Izzuddin Bin Mohd Zulkurnin
Matric Num : 051566
Supervisor : DR. Wan Nor Shuhadah Binti Wan Nik

Background
2
• Cloud Computing system is the most important for any organization to do any task in the
real world today.
• It is related with the computer system performance, that any organization need the best
performance to do any kind of their task.
• It also related about resource utilization which mean CPU utilization, Time-sharing
response, troughput and any kind of system performance.
• This project is focus on the scheduling resource utilization.
• Cloudsim is the simulation tool that is used in the overall implementation Cloudlet
scheduling algorithm.
• Firefly algorithm is the new scheduler algorithm that will be implemented in the
Cloudsim simulator.
• It will be compared with the default algorithm scheduler and at the end of this project is to
analyze which algorithm is the best algorithm for scheduling

Problem Statement
• To find the optimal value CPU utilization to prevent the computer processor from
idle and overheating.
• The wrong Allocation of job Scheduling does affect the performance of the computer
due to unable to create a parallel and synchronized process that occurs at a time.
• The inflexibility of scheduling algorithms will cause longer processing time which
further degrades system performance. which had constraints on the scheduling
algorithm.
3

Objective
• To study the resource scheduling algorithm using a cloudsim simulator
• To implement Firefly scheduling algorithm into the cloud computing simulator in
achieving optimal resource utilization.
• To analyze the performance of one of the latest metaheuristic algorithm (i.e. Firefly
algorithm) for scheduling problems using Cloudsim simulator and compare to other
traditional scheduling algorithms (i.e. FCFS ) in terms of resource utilization.
4

Scope
• Analyze new scheduling algorithm (i.e. Firefly algorithm ) against a traditional
algorithm (i.e. FCFS) implemented in Cloudsim simulator.
• Records resource utilization by plotting some graphs and identify better scheduling
algorithm in achieving an optimal resource utilization.
5

Academic Value
• The process of study and understanding of the Cloudsim architecture.
• The process of study on a new Scheduling algorithm like i.e. firefly algorithm as a
scheduling algorithm that has been implemented in the Cloudsim simulator.
6

Limitation
• Limited time to study the whole concept of the Firefly algorithm to be implemented
in the Cloudsim simulator.
• The implementation of a scheduling algorithm in the simulation environment in real-
time computer performance cannot be similar to simulation performance.
• The implementation of scheduling in the real world is difficult and costly to do.
7

Literature Review
 What is Cloudsim?
• Cloudsim is a extensible simulation framework[1].
• Enables for modelling, simulation, and experimentation of emerging Cloud
Computing infrastructure and application services[1].
• Allow cloud developers to test and analyze the performance of cloud computing
resources.
8

Literature Review (1)
 Cloudsim Feature[2]:
• Support for modeling and simulation of Large scale Cloud Computing data centers.
• Support for Modeling and simulation of energy-aware computational resources.
• Support for modeling and simulation of application containers.
• Support for data center network topologies and message-passing applications.
• Support for user-defined policies for allocation of host to virtual machines and policies
for allocation of host resources to virtual machines9

10
Gridsim:
- Grid services
- Core Element
3
Cloudsim:
- User Interface Structure
- Virtual Machine Services
- Cloud Services
- Cloud Resources
2
Usercode:
- Simulation Specification
- Scheduling Policy
1
Simjava:
- Discrete Event Simulation
4
Clouds Architecture Layered[3]:

Cloudsim Architecture Layered[3]:
11
 User code :
• User Code that exposes
configuration related
functionalities for host,
applications, VMs, number of
users and their application types,
and broker scheduling policies.
 Cloudsim :
• Implemented by
programmatically.
• Provided for modeling and
simulation
• Capable of simultaneously
instantiating transparently
and managing a large scale
Cloud framework

Algorithm/ Technique/
Appraoches Advantages Disadvantages
FCFS
• Doesn’t include any complex logic
• Simple and easy to implement.
Starvation doesn’t occur
(every process get a chance to run)
• No option for pre-emption of a
process
• The next process wait for along
time to be executed their
process
SJF
• short processes are executed first
• Troughput increase (more process
can be executed in less amount of
time)
• Longer process
• More waiting time
• Starvation occurs
12
 Comparison of CPU Scheduling Algorithm/ Technique and Approaches:

Appraoches
Advantages Disadvantages
Round Robin (RR)
• Starvation doesn’t occur(every
process is given a fixed time to
execute)
• No process is left behind
• If time quantum is shorter than
needed,
• Number of times that CPU
switches from one process to
another process, increases.
This leads to decrease in CPU
efficiency.
Priority based Scheduling
• Priority of process selected based on
memory requirement, time
requirement or user preference
• Second scheduling algorithm
is required to schedule the
process which have same
priority
• Starvation occurs13

Approaches
Highest Response Ratio
Next (HRRN)
• Improve both resource utilization
and job response time.
• Unable to overcome the server
overflow problem
Modeling a MapReduce
job using Stochastic
values.
• Achieve maximum utilization of
resources.
• Difficult to allocate resources
in a mutually optimal way due
to the lack of information
sharing between them14

15
Load Balancing System
Static Load Balancing Dynamic Load Balancing
• Do not modify or
change state at
run time.
• Do not play a big
role
• Definitely modify and
changes state at run
time
• Play a big role in
cloud environment
What is Load Balancing ?
• Acts as the traffic cop.
• Sitting in front of yours server and
routing protocol.
• In a manner, maximize speed and
capacity utilization.
• Ensured that no one server is
overworked.

16
More predictable Susceptible to fault
Less resource utilization Low efficiency
Easy to design and implement No accurate method to estimate execution time
Static Load Balancing System

17
Dynamic Load Balancing System
Aware of the system and high adaptability Less predictability
Fault tolerant Uses high resources
Easy to estimate execution time Algorithm are complicated

Expectation Result
18
 Three performance metrics are measured and presented[4]:
• Makespan - Completion of all the cloudlets execution in workload.
- The smaller value of Makespan represents a better execution performance.
• Throughput - Number of jobs executed during the span of per unit time
- Number of cloudlets executed per second
• Average Resource Utilization Ratio (ARUR) - Ratio of average Makespan to the Makespan of the
cloud System
- ARUR value remain between 0 and 1 where value
close to 1 shows exceptional resource utilization.

Expectation Result
19
Composition of benchmark GoCJ dataset[4] Composition of synthetic workload
Three Dataset:
 GoCJ dataset?
• Serve as an alternative to benchmark
workload for scheduling and
resource-allocation mechanism

Expectation Result
20
 Example of Makespan result with three dataset :

Expectation Result
21
 Example of Average makespan result with three dataset :

Expectation Result
22
 Example of throughput result with three dataset :

Expectation Result
23
 Example of Average throughput result with three dataset :

Expectation Result
24
 Example of ARUR result with three dataset :

Expectation Result
25
 Example of Mean ARUR result with three dataset :

Expectation Result
26
 Percentage of resource utilization of scheduling algorithm :

Framework of Resources Scheduling
• A workload management system is where the process
management for job scheduling is located in cloud computing.
• Two sections that communicate to allocate resources in the cloud
computing system.
• The sections included are resources provisionor and resources scheduler.
• Resources provisionor allows the allocation of a cloud provider’s resources
to a customer.
• Resources scheduler is where it assigns the resources they have to jobs, tasks or projects they need to
complete, and schedule start and end date for each task or project based on resource availability.
28

Cloud Computing Scheduling Model
• Broker is the middle interface between the client and the resource provider.
• Broker it plays the main role in the process of the resource scheduling process.
• Firstly the client submits the task to the broker.
• Then broker searches for the resources in information service and then deploys the
task to the appropriate resources according to the algorithm provided to the broker.
• Broker contains Job Control Agents, Scheduler advisor, Explorer,
Trade Manager and Deployment Agent.
29

(Continued)
• Job Control Agent that takes charge in monitor jobs that happen in software system
such as when the schedule is generated, how the job is creation happens, jobs status
and communicate with client as well with schedule advisor.
• Schedule advisor that is used in order to determine resources, allocate available
resources.
• Cloud Explorer which communicates with cloud information services to search for
resources and record resources status information as well as identifies the list of authorized machines.
30

(Continued)
• Trade Manager that determines the cost of resources access and makes an attempt to
communicate with resources at a low cost under the guidance of schedule advisor.
• Deployment Agent that uses scheduler instruction to activate the execution of tasks.
It also updates the status of execution and sending back to Job Control Agent at
regular intervals.
31

Cloudsim Model
32
• Data centers : the resource provider, includes one or more hosts.
• Host: the physical machine that allocates one or more VMs.
• Virtual machine: machines on which the cloudlet will be executed.
• Cloud Information Service (CIS): responsible for registering all resources of data centers.
• Broker: when a broker has the DC characteristics, it will submit VMs to the specific host in the specific DC and then allocate the
cloudlets (tasks) to specific VMs. Finally, the broker will destroy the free VMs after execution of all cloudlets.
• Cloudlets: In Cloudsim, cloudlets are all tasks and applications that are executed on VMs.

Cloudsim Working Model
33
Step 1:
- To start with CIS object will be created and it gets registered
- I.e cloud information service
Step 2:
- Next step is the creation of data center.
- Data center contains Hosts which are physical and host has configuration like Number of Processing elements (PEs), Bandwidth,
Ram ,harddisk, Number of CPU’s etc.
- When the data center is created then it is registered into the CIS .

34
Step 3:
- Create the broker object
- It queries the CIS for the data center.
- Broker requires to submit the tasks to the data center.
Step 4:
- CIS returns broker the registered data centers into the CIS with their configuration.
Step 5:
- Cloudlets are submitted to the broker for its execution. E.g. file input size.

35
Step 6:
- Broker in turn submit the cloudlets to the VM’s according to the various policies
(which I will be posting soon) which runs on the physical machines
in the data centers which in turn uses the physical resources.

Basic steps in java code from
configuration to simulation
36
1-Setting the number of users for a current simulation.
2-Initializing the simulation by instantiating the common variables (current time, trace flag, number of users).
3-Creating CIS instance.
4-Creating data center instance and then registering it with CIS.
5-Creating physical machines (hosts) with their characteristics.
6-Creating data center broker instance.

Basic steps in java code from
configuration to simulation
37
7-Creating VMs with their characteristics.
8-Submitting VMs to data center broker.
9-Creating cloudlets and specifying their characteristics.
10-Submitting cloudlets to data center broker.
11-Sending a call to start the simulation once there is an event to be executed.
12-Sending a call to stop the simulation once there is no event to be executed.
13-Printing the results of the simulation.

Design of Cloudsim
38
 There are several of classes need to modify if want to implement own algorithms:
• DatacenterBroker - Modifying the way VM provisioning requests are submitted
to data centers and the way cloudlets are submitted and assigned to VMs.
• VmAllocatonPolicy - Need to extend this abstract class to implement your own
algorithms for deciding which host a new VM should be placed on.
• VmScheduler - Implementing algorithms for resource allocation to VMs within a
single host.
• CloudletScheduler - Implementing algorithms for scheduling cloudlets within a
single VM.

First Come First Serve Algorithm
• FCFS is one of the process scheduling algorithms that commonly used in task scheduling of the operating systems.
• FCFS is a mechanism that executes a request in a queue and processes it based on their arrival order.
• First-come will be handled first and the next job will be executed once the previous job is complete.
• FCFS is the scheduling able to provide efficiency.
• FCFS can easily be applied since it is a simple scheduling algorithm.
40

Pseudocode of FCFS
42
1. Firstly start the program.
2. Read the number of processes count in (int).
3. Read the burst time of all processes from the user.
4. Read the arrival time of each process for this scheduling.
5. Calculate the waiting time and turn around time.
6. Display it coming the result of first come first serve Scheduling.
7. Calculate average waiting time and average total turn-around time and display it on screen.
8. Terminate the program.

Java sourcecode of FCFS
(continued)
44
The Output:

Firefly Algorithm
45
• Firefly algorithm (FA) is a metaheuristic algorithm.
• Inspired by the flashing behavior of fireflies.
• Attractiveness is proportional to their brightness.
• Brightness can decrease as their distance increases.
• Two important variables which is the light intensity and attractiveness.
• The light and attractiveness are decreased as the distance increase.

Firefly Algorithm (continued)
46
 Formula of Light Intensity:
Symbol Meaning
I Light intensity
I0 Light intensity at the initial or original light
intensity
y The light absorption coefficient
r Distance between i and j
Where,

Firefly Algorithm (continued)
47
 Formula of Attractiveness:
Where,
Bo Attractiveness at r is 0
 Formula distance between firefly:

Flowchart of Firefly Algorithm
48

Pseudocode of Firefly Algorithm
49

Ganchart
50
NO ACTIVITIES WEEK
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
1 Final Year Project I briefing
1.1 Topic Discussion and Determination
2 Project Title Proposal
2.1 Proposal Writing (Chapter-1 Introduction)
2.2 Proposal Writing (Chapter-2 Literature Review)
2.3 Proposal Writing (Continued)
3 Proposal Progress Presentation and Panel's Evaluation
4 Methodology Workshop
SEMESTER BREAK
4.1 Proposal Writing (Chapter 3 - Methodology Proof of Concept (POC)
5 Final Year Project Format Writing Workshop
5.1 Drafting Report of Proposal
6 Submit Draft of Report to Supervisor
7 Preparation for Final Presentation
8 Final Presentation and Panel's Evaluation
9 Final Report Submission and Supervisor's Evaluation

References
[1] Soni,Mitesh. (2014, March 3). The CloudSim Framework: Modelling and Simulating
the Cloud. Retrieved from https://opensourceforu.com/2014/03/cloudsim-
framework-modelling-simulating-cloud-environment/.
[2] http://www.cloudbus.org/cloudsim/
[3] Khatibi, arezoo. (2018, June 14). GroundAI. Retrieved from
https://www.groundai.com/project/criteria-for-the-cloudsim-environment/1.
[4] Hussain, A.—Aleem, M.: GoCJ: Google Cloud Jobs Dataset for Distributed and Cloud
Computing Infrastructures. MDPI Data, Vol. 3, 2018, No. 4, pp. 1–12, doi:
10.3390/data3040038
51

Task Scheduling Using Firefly algorithm with cloudsim

More Related Content

What's hot (20)

Similar to Task Scheduling Using Firefly algorithm with cloudsim (20)

Recently uploaded (20)

Task Scheduling Using Firefly algorithm with cloudsim