The Amazon cloud provides infrastructure as a service (IaaS), It provides computing infrastructure such as for servers, storage or network end points of a desired capacity are virtually provided in minutes through an automated web-based management console.

This core IaaS service, called Elastic Compute Cloud, or EC2, EC2 is also often used to describe the entire cloud offering

Cloud users have root/administrator access to these servers, and therefore control them completely.

For example, they can deploy applications and make them publicly accessible over the internet.

Static network addresses required for such purposes can also be provisioned through the management console.

Thus, VM4 is also accessible by internet users at large over HTTP. Such publicly available static IP addresses are charged on a fixed monthly basis

Network data transfer to any server, with or without a static IP address, is charged on usage basis. Users can provision and access many servers that can communicate with each other over the fast internal network within the Amazon cloud.

If VM4 is a web server, VM5 may be a database server, and these two communicate over the internal cloud network using TCP/IP.

Since VM5 is a database server, it needs to store and retrieve data.

The Amazon SimpleDB service provides an object store where key-value pairs can be efficiently stored and retrieved.

It is important to note that SimpleDB is not a relational database, and we shall describe the features provided by such key-value pair cloud databases in a later section.

Instead of using SimpleDB, virtual servers could instead use a relational database system, which may come either pre-installed as part of the AMI, or separately by users in the normal manner. However, it is important to understand that virtual servers do not have any persistent storage; so any user data on file system is lost when the server shuts down. In order to store data persistently, such in a relational database,

Elastic Block Storage needs to be mounted on a virtual server.

The Elastic Block Storage service maintains persistent data across all users on a large set of physical servers.

After a virtual server boots, it must attach user data from the EBS as a logical storage volume mounted as a raw device (disk).

VM6 might be an archival server where VM5 sends logs of whatever updates it makes to the SimbleDBdatastore. VM5 sends data to VM6 not over TCP/IP, but using the Amazon Simple Queue Service (SQS).

The SQS is a reliable persistent message queue that is useful for temporarily storing data that needsto eventually get to a processing server such as VM6,

The Amazon S3 Storage Service provides a different storage model. Data in S3 can be files of any type, and in general any blob (binary large object).

Users access and modify S3 objects via URIs, using REST web services (which we shall cover in Chapter 7). S3 objects are accessible over the internet as well as from virtual servers within the Amazon cloud.

Storage in S3 is also used for storing machine images (AMIs) that users define themselves.

Runtime performance parameters, such as CPU and I/O utilization, of a user’s virtual servers can be monitored in real-time by Amazon Cloud Watch; this data can be used by Amazon Auto Scale to add (or remove) virtual servers from an application cluster and automatically provision them with predefined machine images. Finally, Elastic Load Balancing allows a group of servers to be configured into a set across which incoming requests are load balanced.

The performance statistics of the load-balanced requests can also be monitored by Cloud Watch and used by Auto Scale to add or remove servers from the load balanced cluster.