Characteristics of a cloud

Assignment

Beat Kläy, the president of the board, writes to you before the next meeting. "I ran servers for a municipality for thirty years. Please explain what is different about a cloud compared with a well-run data center. Keep it short, I want to show it to the board on one page."

Learning goals

After this chapter you can:

• Explain why cloud is an operating model and not a place
• Name the five characteristics of a cloud according to NIST and explain them with examples
• Check whether an offering meets these characteristics
• Argue why the small cloud of Wolkenwerk Bern is a cloud as well

Cloud as an operating model

In everyday language "cloud" sounds like a distant data center run by a large provider. In the field, the term means an operating model, a way of providing and using IT resources. Computing power, storage and network sit in a shared pool. Whoever needs them takes them, uses them as long as needed and gives them back.

A well-run data center can have the same hardware as a cloud. The difference is how you get a resource. In classic operations someone files a request, the operations team orders or sets up a server, and after days or weeks it is ready. In a cloud the user provisions the resource themselves, within minutes, through a web interface or an API.

The five characteristics according to NIST

The most widely cited definition comes from the US National Institute of Standards and Technology (NIST), published in 2011 as Special Publication 800-145. It names five characteristics a cloud has to meet.

Characteristic	Meaning	In the lab
On-demand self-service	Users provision resources themselves, without a person at the provider having to step in.	A member starts a VM from a template in Sunstone (chapter 3.4).
Broad network access	Resources are reachable over the network with standard protocols, from different devices.	Web interface, OpenNebula API, SSH to the VMs
Resource pooling	Resources form a shared pool for several tenants, who do not need to know which hardware their VM runs on.	The scheduler spreads VMs across the nodes, and the members are tenants (chapter 3.1).
Rapid elasticity	Resources grow and shrink quickly with demand, partly on their own.	The web shop scales under load (chapter 5.4), and OneFlow grows services (chapter 3.5).
Measured service	Usage is measured and visible to both provider and users.	Showback per member (chapter 3.3), metrics in Prometheus (chapter 5.1)

If one of the characteristics is missing, it is strictly speaking not a cloud. Even an efficient virtualization platform is not a cloud if only the operations team creates VMs on it, because self-service is missing.

Size and ownership

The definition says nothing about size and nothing about who owns the hardware. The three servers of Wolkenwerk Bern form a cloud as soon as the members provision resources themselves, several members share the pool, usage is measured and services grow with the load. You build these characteristics one at a time during the course. Part 2 sets up the pool and the network access, part 3 adds self-service and measurement, and part 5 adds elasticity when the web shop grows under load. In the advanced tasks of part 3 a service already grows with OneFlow.

Exercise Classifying

Match each statement to one of the five characteristics.

1. "Treuhand Aaregg sees every month how many CPU hours it used."
2. "Praxis Rosenweg starts a new VM on a Sunday evening without calling anyone."
3. "On Saturday the web shop runs eight copies instead of two."
4. "The municipality does not know which node its VM runs on, and it does not need to."
5. "The members reach their VMs from the office and from home."

Solution

1. Measured service
2. On-demand self-service
3. Rapid elasticity
4. Resource pooling
5. Broad network access

Exercise Briefing

Write five sentences for Beat Kläy, one per characteristic, explaining what sets Wolkenwerk Bern apart from a classic data center.

Solution

Members get their servers themselves and within minutes instead of through a request. They reach everything over the network with standard protocols. The three servers form one pool that all members share. Whoever needs more capacity gets it at short notice and gives it back afterwards. Every member sees its usage and pays accordingly.