Qualitative and quantitative research on the availability of cloud computing (4)

"China Cloud net Exclusive" Chen Whilin, the chief consultant of Cloud Network

4 Case Study-Amazon AWS

4.1 Amazon AWS System Architecture

Amazon AWS was online in 2006 and is an IaaS in cloud computing services (infrastructure as a service). Amazon AWS provides data center clusters in various regions of the world (Region). It is divided into 4 major regions. These include the North American region (northern Amerian Region), the Asia Pacific region (Asia and Pacific Region), Europe (including the Middle East and Africa) (Europe/middle East/africa) and the South American region ( South America Region), as shown in Figure 5.

Figure 5 Amazon Data center group distribution map

Here are the specific distributions of the 4 largest regions of the Amazon World:

North America Large Area:

Divided into 5 regions:

US East (Northern Virginia) Region EC2 availability zones:5 2006 start service

US West (Northern California) Region EC2 availability zones:3 2009 Start Service

US West (Oregon) Region EC2 availability zones:3 2011 Start Service

AWS Govcloud (US) Region EC2 availability zones:2 2011 Start Service

(Note: Govcloud is a cloud-computing area that Amazon is carrying separately for U.S. government agencies.) in the northwest of the United States

Asia-Pacific region

Asia Pacific (Singapore) Region EC2 availability zones:2 2010 Start Service

Asia Pacific (Tokyo) Region EC2 availability zones:3 2011 Start Service

Asia Pacific (Sydney) Region EC2 availability zones:2 2012 Start Service

European Region

EU (Ireland) Region EC2 availability Zones:3 started service in 2007

South America Region

São Paulo Region EC2 availability zones:2 2011 Start Service

In addition, Amazon in order to improve the reliability of the business, stability and user performance experience, but also around the world to establish a CDN Edge Acceleration Node Data center (Edge Location). The distribution area, shown in Figure 6, is built around the global area and the available area nodes, which speeds up user access to data.

Figure 6 Amazon Edge location distribution map

In order to maximize the reliability of the IaaS service, each zone contains multiple EC2 available areas (availability Zone). The available zones correspond to dispersed physical data centers that are exclusive to one region (Region). which,

* A region can contain multiple available areas;

* A usable area can only belong to one zone.

In relation to the area and available areas of AWS, Figure 7 shows a relatively clear topological relationship.

Figure 7 AWS area and available area

The different regions are completely independent and there is no information interaction. thereby removing one area failure failure can bring influence on another area.

The available areas of the same zone are separate data centers. The expiration of one free zone does not affect other available areas within the same zone. In the same area, multiple available areas are interconnected through high-speed, low-latency networks.

Users can choose from their own business needs of customers closer to the region, such as the East or west of the United States.

Users can also avoid single point failures in the available zones by starting multiple businesses in different free zones within the same zone.

In one usable area, an AWS service consists of multiple parts and service module instances (Instance). As shown in Figure 8, AWS is composed mainly of EC2 (elastic Compute Cloud), EBS (Elastic block Store) and S3 (Simple Storage Service). Where EC2 provides virtualized server host resources; EBS provides block device support for EC2, similar to a large hard drive. Then it can provide various file system support for EC2 by format. S3 provides a data storage and backup resource that can be accessed via the Web, which allows users to perform real-time or periodic backups of the data generated in the AWS EC2 or EBS, and can be accessed via APIs and various web interfaces.

Figure 8 AWS Overall Structure chart

Elb (elastic load balancer) is an optional load-balancing component. With Elb, an AWS tenant can deploy multiple instances of EC2 in an available area of AWS or in multiple available zones. Then, the purpose of load balancing and fault-tolerant is realized by Elb. RDS (relational database service) is a Web interface based relational database services, and AWS tenants can generate and manage their own database business through RDS's management interface. The database can be mysql,oracle or Microsoft SQL Server. CloudFront is a web-based CDN acceleration service in AWS that allows tenants to easily manage content and decide on various acceleration options for content, or with other service modules in AWS, such as S3 for data access acceleration.  The VPC (Virtual private Cloud) service of AWS is a solution for tenants to build their own private networks in the AWS Cloud. Similar to the category of hosted Private clouds. A tenant can create a VPC in AWS and then fully control the two-tier, three-tier configuration of the network in this VPC domain, and various security controls.

Broadly speaking, a basic AWS cloud service consists of EC2, EBS, RDS and S3. Other services are optional, such as ELB for load balancing and fault tolerance.

(Responsible editor: Lu Guang)