An approach for bandwidth guarantee in multi-tenant cloud datacenter network

Other Title(s)

منهجية ضمان عرض النطاق في شبكات مراكز البيانات متعددة المستخدمين

Dissertant

Barud, Ayman Ismail Khalil

Thesis advisor

Barakah, Ribhi Sulayman

Comitee Members

Mahmud, Ahmad Yahya
Barhum, Tawfiq Sulayman

University

Islamic University

Faculty

Faculty of Information Technology

University Country

Palestine (Gaza Strip)

Degree

Master

Degree Date

2015

English Abstract

The rapid development in cloud infrastructure allows cloud providers to offer computing power such as CPU, memory and storage to different customers, and serve multiple tenants at the same time.

Unfortunately, due to network nature, cloud provider give no guarantee of network resource allocation to individual VM, relying on the fact that VMs are unlikely to simultaneously maximize their use of their nominally assigned bandwidth.

However, many research shows that there is potential for VM to effect each other due to contention on network resource, or as side effect of noisy neighbor, and varying of bandwidth demand across time.

Resulting in a severe lack of predictability and fairness of application performances.

That cost tenants unpredictable cost and provider loss revenue.

Running cloud application without deprecation of performance require cloud provider to offer both computation power and network resource as bundle.

The new approach that combine both computing and network resource in one form called virtual data center (VDC), formally known as VDC embedding problem.

In this approach, cloud provider should provide tenants with an interface to explicit specifying their bandwidth demand.

Cloud provider achieve deterministic bandwidth guarantee by proper placement of tenant VMs and static enforcement of rate limit.

However, offering such service to customer is not easy task, and many researchers classify this problem as multiple dimension ben packing problem, which been as NP hard problem.

This research contribute to work done in this area.

We introduce two stage heuristic algorithms.

The first stage, preprocess tenant VDC request by consolidating VMs and link capacity.

The second stage, we map tenants request to physical datacenter server and links.

We evaluate our approach against other approach using cloud computing simulation tools, and find that our approach increase the utilization of network, while maximizing the number of accepting VDC request, in addition to provide some level of reliability to tenant request against server failure.

Main Subjects

Information Technology and Computer Science

No. of Pages

61

Table of Contents

• APA
• MLA
• AMA

Language

English

Data Type

Arab Theses

Record ID

BIM-688652