Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                
Scribd
Upload
5 views

Dynamic Optimization of Multiattribute

The document presents a dynamic optimization scheme for resource allocation in self-organizing clouds (SOC) using a VM-multiplexing approach that enhances resource utilization and execution efficiency. It introduces a lightweight multi-attribute range query protocol that minimizes message size and contention, achieving a 15-60% improvement in system throughput over traditional P2P Grid models. The proposed system addresses challenges in decentralized environments, ensuring adaptability and efficient resource provisioning while maintaining quality of service within user budgets.

Uploaded by

vinoth
Copyright
© © All Rights Reserved
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
5 views

Dynamic Optimization of Multiattribute

The document presents a dynamic optimization scheme for resource allocation in self-organizing clouds (SOC) using a VM-multiplexing approach that enhances resource utilization and execution efficiency. It introduces a lightweight multi-attribute range query protocol that minimizes message size and contention, achieving a 15-60% improvement in system throughput over traditional P2P Grid models. The proposed system addresses challenges in decentralized environments, ensuring adaptability and efficient resource provisioning while maintaining quality of service within user budgets.

Uploaded by

vinoth
Copyright
© © All Rights Reserved
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
You are on page 1/ 14

Dynamic Optimization of Multiattribute

Resource Allocation in Self-Organizing


Clouds
Abstract:

VM-multiplexing resource allocation scheme to manage decentralized resources. Our


approach not only achieves maximized resource utilization using the proportional share model
(PSM), but also delivers provably and adaptively optimal execution efficiency. We also design a
novel multi-attribute range query protocol for locating qualified nodes. Contrary to existing
solutions which often generate bulky messages per request, our protocol produces only one
lightweight query message per task on the Content Addressable Network (CAN). It works
effectively to find for each task its qualified resources under a randomized policy that mitigates
the contention among requesters. We show the SOC with our optimized algorithms can make an
improvement by 15-60 percent in system throughput than a P2P Grid model. Our solution also
exhibits fairly high adaptability in a dynamic node-churning environment.

Existing System:

Cloud architectures are not without problems. Most cloud services built on top of a
centralized architecture may suffer denial-of-service (DoS) attacks, unexpected outages, and
limited pooling of computational resources. On the contrary, volunteer computing systems (or
Desktop Grids) can easily aggregate huge potential. Computing power to tackle grand challenge
science problems, by leveraging virtual machine (VM) technology which provides performance
and fault isolation, cloud resources can be provisioned on demand in a fine grained, multiplexed
manner rather than in monolithic pieces. By integrating volunteer computing into cloud
architectures, we envision a gigantic self-organizing cloud (SOC) being formed to reap the huge
potential of untapped commodity computing power over the Internet.
Disadvantage:

1) The multiattribute range query problem in a fully decentralized environment for locating a
qualified node to satisfy a user task’s resource demand with bounded delay.
2) Optimize a task’s execution time by determining the optimal shares of the multiattribute
resources to allocate to the tasks with various QoS constraints, such as the expected execution
time and limited budget.

Proposed System:
VM-multiplexing resource allocation scheme to manage decentralized resources. Our
approach not only achieves maximized resource utilization using the proportional share model
(PSM), but also delivers provably and adaptively optimal execution efficiency. We also design a
novel multiattribute range query protocol for locating qualified nodes. Contrary to existing
solutions which often generate bulky messages per request, our protocol produces only one
lightweight query message per task on the Content Addressable Network (CAN). It works
effectively to find for each task its qualified resources under a randomized policy that mitigates
the contention among requesters.

This paper proposes a novel scheme (DOPS) for virtual resource allocation on a SOC, with three
key contributions listed below.

1. Optimization of task’s resource allocation under user’s budget: we propose a solution


which can optimize the task execution performance based on its assigned resources under
the user budget.
2. Maximized resource utilization based on PSM: In order to further make use of the idle
resources, we design a dynamic algorithm. This can give incentives to users by gaining
an extra share of unused resource without more payment.
3. Lightweight resource query protocol with low contention.
Advantage:

 The SOC with our optimized algorithms can make an improvement by 15-60 percent in
system throughput than a P2P Grid model.
 Exhibits fairly high adaptability in a dynamic node-churning environment.
 Self-organizing cloud (SOC), which can connect a large number of desktop computers on
the Internet by a P2P network.

SYSTEM REQUIREMENTS:
SOFTWARE REQUIREMENTS:

• Front End : Java

• Environment : Eclipse/Net Beans

• Back End : SQL Server 2005

• Operating System : Windows XP

HARDWARE REQUIREMENTS:

• Processor : Pentium IV

• RAM : 512 MB

• Hard Disk : 80 GB

Problem Analysis

Resource allocation problem in cloud systems emphasizes how to


harness the multiattribute resources by multiplexing operating systems.
In recent years, various enhancements on resource isolation techniques
have been proposed to achieve fine-grained dynamic resource
provisioning. But the solutions create an unprecedented opportunity to
maximize resource utilization, which were not possibly applied in most
Grid systems that usually treat the underlying resources as indivisible
ones and prevent simultaneous access to them. Today’s cloud
architectures are not without problems. Most cloud services built on top
of a centralized architecture may suffer denial-of-service (DoS) attacks ,
unexpected outages, and limited pooling of computational resources. On
the contrary, volunteer computing systems (or Desktop Grids) can easily
aggregate huge potential computing power to tackle grand challenge
science problems.

Modules:

Resource Provisioning in SOC Clouds:


In modern virtualization based soc clouds, applications share the underlying hardware by
running in isolated Virtual Machines (VMs). Each VM, during its initial creation, is configured
with a certain amount of computing resources (such as CPU, memory and I/O). A key factor for
achieving economies of scale in a soc cloud is resource provisioning, which refers to allocating
resources to VMs to match their workload. Typically, efficient provisioning is achieved by
dynamic resource provisioning. VM capacity is dynamically adjusted to match workload
fluctuations.

Random route query messaging:


The proposed PG-CAN range query protocol in this work aims to find the qualified
resources with minimized contention among requesters based on task’s demand. It is unique in
that for each task, there is only one query message propagated in the network during the entire
course of discovery. This is different from most existing multiattribute range query solutions that
require to propagate multiple subqueries along multiple dimensions in parallel. To mitigate the
contention problem due to analogous queries in CAN, our range query protocol proactively
diffuses resource indexes over the network and randomly route query messages among nodes to
locate qualified ones that satisfy tasks’ minimal demands.
Load Distribution:
We distributed incoming task to available system resources and achieving good load
balance in a fully decentralized and heterogeneous cloud environment. Allocate resource for task
with its resource requirements that can minimize a task’s execution time. We investigate three
different policies to control imported tasks or disperse the load distribution, namely double-
check policy, queue-assistant policy, and extra-virtual-dimension policy.
Minimizing Task Execution Time:
Dynamic optimal proportional-share resource allocation method, which leverages the
proportional share model. The key idea to redistribute available resources among running tasks
dynamically, such that these tasks could use up the maximum capacity of each resource in a
node, while each task’s execution time can be further minimized in a fair way. DOPS consists of
two main procedures: 1) Slice handler: It is activated periodically to equally scale the amount of
resources allocated to tasks, such that each running task can acquire additional resources
proportional to their demand along each resource dimension. 2) Event handler: It is responsible
for resource redistribution upon the events of task arrival and completion.
Fault Tolerance:
Cloud systems usually do not provision physical hosts directly to users. If the resources
provisioned are relatively sufficient, we can guarantee task’s execution time always within its
deadline even under the wrong prediction about task’s workload. Each task can be guaranteed to
be finished within its original deadline even though task properties cannot be predicted
accurately.

Algorithm 1: Deadline-driven provisioning


1. Foreach request with QoS constraints do
2. Resources available resources for the application;
3. PendingT asks number of tasks in the queue;
4. Eft = ( pendingT asks * averageTaskRuntime) / (resources);
5. If eft > applicationT imeRemaining then
6. extraResources [(pendingT asks * averageT askRuntime) /
(applicationT imeRemaining)];
7. Provisioner.selectResources(applicationId,extraResources);
8. else
9. toRelease 0;
10. If pendingTasks < resources then
11. toRelease (pendingT asks – resources);
12. END
13. Else
14. pendingT asks (pendingT asks + runningT asks);
15. Eft = ( pendingT asks * averageTaskRuntime) /(resources);
16. If eft > applicationT imeRemaining
17. toRelease resources – [(pendingT asks *
averageTaskRuntime) /(applicationT
imeRemaining)];
18. END
19. END
20. Provisioner.releaseResources(applicationId,toRelease);
21. END
22. END
Architecture:
Uml Diagrams:

Use Case Diagram:


Class Diagram:

User Node
name name
Password Task Analysis ID VMM
Port Resource Attribute
IP Expected Resource()
Managing nodes()
Task workload() Registration()
Registration() Qualified node Checking()
Task submission() Task operation()

Sequence Diagram:

User Registration Node VMM

Register or Sign in

Register or Sign in

Check and insert Managing Nodes

Task Submission

Communication With Neighbors

Task Result
Activity Diagram:

Start

User Node

Register

Check
No

Yes

Task
Submission

Task
Scheduling

Qualified Node
Selection

Execute Task
Conclusion:
We propose an algorithm to optimize the task execution time on a qualified resource
node, given its preset budget and tolerable quality of service. We further propose a dynamic
optimal proportional-share (DOPS) resource allocation algorithm. The key idea is to dynamically
scale the amount of resources at each dimension among running tasks proportional to their
demand, such that these tasks could use up the maximum capacity of each resource type at a
node. We also design a novel multiattribute range query protocol for locating qualified nodes.
Contrary to existing solutions which often generate bulky messages per request, our protocol
produces only one lightweight query message per task on the Content Addressable Network
(CAN). It works effectively to find for each task its qualified resources under a randomized
policy that mitigates the contention among requesters.

You might also like