Cloud Computing Networks: Utilizing the Content Delivery Network

Cloud Computing Networks: Utilizing the Content Delivery Network

Yale Li (Microsoft Corporation, USA), Yushi Shen (Microsoft Corporation, USA) and Yudong Liu (Western Washington University, USA)
DOI: 10.4018/978-1-4666-4801-2.ch010
OnDemand PDF Download:
$30.00
List Price: $37.50

Abstract

Cloud Computing has the potential to trigger a major computing model transformation for the IT industry. This chapter briefly describes the business and technical benefits of Cloud Computing and explains the technical challenges in Cloud Computing, such as the network bottleneck. One of the solutions to address the network problem is the Content Delivery Network (CDN). Here, the basics of the Akamai CDN technology is digested. Then, the authors conduct a CDN experiment in the Microsoft public cloud, Windows Azure, to demonstrate the benefits of CDN integration with the cloud. The results show significant gain in large data download by the utilization of a CDN. Finally, a couple of academic research ideas are summarized for future improvements on the CDN model.
Chapter Preview
Top

Opportunities And Challenges Of Cloud Computing

Migration from client/server computing to cloud computing is a major computing model transformation. There are great opportunities for both business and technical innovations. However, tremendous challenges are also present.

Business Opportunities and Challenges

On the business side, cloud computing is to make the business more agile, by utilizing resources more effectively to achieve lower cost of ownership. From the cloud provider’s perspective (Armbrust & et. al., 2009), the following formula is used, to describe the profit model for cloud service providers:

(1)

The left-hand side represents the expected profit from using cloud computing. The right-hand side represents the expected profit from using the traditional data center. Both sides perform the same calculation by multiplying profits per user-hour by the total user-hours.

Because the resource utilization in cloud computing is 100% (fully utilized), and the resource utilization in the traditional data center is less than or equal to 100% (fully utilized), the true cost of the traditional data center is greater than or equal to the cost of cloud computing. This formula reveals that cloud providers have a better control over the cost per user-hour, and therefore a better opportunity to gain more profit.

From the cloud customer’s perspective, applications or services with the following workload patterns are to enjoy great benefits by the adoption of cloud services:

  • Unpredictable Bursting: An event may trigger heavy usage of resources: normally, the customer would have to scale the design considerations to try and predict what this resource usage requirement could be;

  • Predictable Bursting: Using Dominos Pizza as an example, the store is very busy on Friday nights. In most days of the week, demand is much less. Even though the additional load is to be maintained, it is expensive for this extra capacity, because it is under-utilized when demand is lower;

  • On and Off: Similar to Predictable Bursting, On/Off can have seasonal or time-bounded workloads where it is either all or almost nothing processing requirements. Important enterprise workloads that are run monthly, quarterly and annually exhibit this type of behaviour;

  • Fast Growing: This is interesting in the case of smaller start-up companies or groups in larger companies. It can also be associated with new development. How to plan for rapid capacity increase is no longer an issue in the cloud, both during development and operations. Elasticity can be a huge opportunity for savings.

Figure 1 demonstrates the above 4 application workload patterns.

Figure 1.

Cloud application workload patterns

The business challenges for cloud adoption is similar to IT outsourcing. It is critical to understand how the availability, security and supportability requirements are to be met by the cloud service providers. It is also important to know the responsibility of each party, and have a formal service level agreement (SLA) and legal contract between the customer and the provider.

Complete Chapter List

Search this Book:
Reset