Understanding Cloud Characteristics - Deploying and Managing a Cloud Infrastructure - page 2

Thomas J. Watson, the founder of IBM, remarked in the
early 1940s, “I think there is a world market for about five
computers.”
Even though that comment was referring to a new line of “scientific” computers that
IBM built and wanted to sell throughout the United States, in the context of the cloud, the
idea behind it still applies. If you think about it, most of the world's critical business infra-
structure relies on a handful of massive—really massive—data centers spread across the
world. Cloud computing has come a long way, from early mainframes to today's massive
server farms powering all kinds of applications.
This chapter starts off with overview of some of the key concepts in cloud computing.
Broadly, the standard features of a cloud are categorized into compute, storage, and net-
working. Toward the end of the chapter, there's a dedicated section on elastic, object-based
storage and how it has enabled enterprises to store and process big data on the cloud.
Basic Terms and Characteristics
Before we begin, it's important to understand the basic terms that will be used throughout
the topic and are fundamental to cloud computing. The following sections will touch upon
these terms to give a feel for what's to follow in later chapters.
Elasticity
Natural clouds are indeed elastic, expanding and contracting based on the force of the
winds carrying them. The cloud is similarly elastic, expanding and shrinking based on
resource usage and cloud tenant resource demands. The physical resources (computing,
storage, networking, etc.) deployed within the data center or across data centers and
bundled as a single cloud usually do not change that fast. This elastic nature, therefore, is
something that is built into the cloud at the software stack level, not the hardware.
The classic promise of the cloud is to make compute resources available on demand, which
means that theoretically, a cloud should be able to scale as a business grows and shrink as the
demand diminishes. Consider here, for example, Amazon.com during Black Friday. There's
a spike in inbound traffic, which translates into more memory consumption, increased net-
work density, and increased compute resource utilization. If Amazon.com had, let's say, 5
servers and each server could handle up to 100 users at a time, the whole deployment would