It’s all the rage and everyone is talking about it. Most small businesses, enterprises, operators and service providers seem determined to leverage it. But what’s all the buzz about? It’s all about something called the cloud, writes Jeff Edlund. Co-sponsored feature: Hewlett-Packard

Jeff Edlund: service providers will be able to try new services to attract new customers and add new application-based revenue streams

In meteorological terms, clouds can come in many varieties. Cloud classifications vary widely from the fluffy white clouds that influence dreams to the awe inspiring Cumulonimbus cloud that soar up into the heavens and almost always bring drenching downpours precipitated by earthshaking thunderclaps.

The technology movement popularly referred to as cloud computing exhibits many similarities to these dramatic atmospheric weather producers. Computer technologists envision many different solution forms that the cloud can take when coalescing into a customer solution.

Additionally, with the current level of hype, almost anything that is already done in IT processing or service delivery can be rained down on users with some sort of cloud label.

Most often these services are referred to as a type of: XaaS or blank as a service. Pick something we all ready do — like computer processing — and it can be given the label of a particular type of infrastructure or service cloud with an acronym like CaaS or compute as a service.

Does this all seem a little familiar?

Remember back in 1984 when the idea of something called X-Windows came out of MIT and was heralded as the emergence of network based computing? Display stations were free to be of any hardware and operating system as long as they had the ability to run the X11 protocol.

Client programs were executed on a central core or data centre computer and the network was used to transfer display information from the application back out to the display station. In 1987 X11 was available to the masses and the era of network based computing was declared to have arrived.

Terminal servers

The year 1985 heralded the emergence of terminal servers. A terminal server is a specialized computer which aggregates multiple communication paths together. Because these paths are bidirectional, two different service types can be supported:

• multiple end points connecting to a single resource;
• a single end point connecting to multiple resources.

Both of these models are widely used. Both physical and virtual resources can be provided through a terminal server. Central computing can provide multiple users access to a remote virtual operating system.

This definition sounds very modern, doesn’t it? With the use of words like “virtual” and “remote” in the definition, it could reasonably be argued by some that because it is both remote and virtual, terminal services are “cloud”.

In 2006 the term “terminal server” got an upgrade and is now used primarily in three ways.

• Console servers — also known as serial terminal servers — are often used for connection to the console ports of Unix servers. This then allows system administrators to connect to the servers over the network.
• A terminal server may refer to a network access server.
• Most commonly, terminal server means a server used in centralised computing.

Today, a client of a terminal server is referred to as a thin client. Often, thin clients are very simple display computers used to present the user with an instant-on web browser that allows the individual to connect to any of a number of applications exposed as simple web-based services.

Is a grid a cloud?

Grid computing started to emerge in the 1990s and is a form of distributed computing where a virtual computer is composed of a cluster or many clusters of networked, loosely coupled computers, acting in concert to perform compute tasks.

What distinguishes grid computing from conventional cluster computing systems is that grids tend to be more loosely coupled, heterogeneous and geographically dispersed.

Also, while a computing grid may be dedicated to a specialised application, it is often constructed with the aid of general-purpose grid software libraries and middleware.

Then the internet tried to grab the cloud

The internet has become a fairly ubiquitous, pervasive, high speed and almost always available network and service delivery resource. With this unparalleled access and robustness, almost any service that is delivered via the internet today can be given the label of “cloud” by its provider.

It is possible, even probable, to declare that if it hadn’t been for the internet, the cloud would have never happened. Clearly, there is still room for standardisation and definition for the concept of cloud.

While X-Windows, terminal services, grid computing and the internet alone are not an exhaustive study of all the events that precipitated cloud computing, they are representative of computing techniques that have been used both past and present to virtualise or abstract some IT function from the user.

Standing alone, none of these early examples are cloud. But perhaps in combination with other well established and emerging computing techniques, a true cloud can be derived.

So can we arrive at a standardised definition of a cloud?

The inhibiting concept in the phrase above is the word “standardised”. If that is the primary criterion then we probably will never arrive at a strict definition of cloud computing. But, it does seem possible that a set of general characteristics can be articulated that will help us recognise what may be a cloud when we see one.

So what set of characteristics should a cloud computing service exhibit?

1: Agility — improves the user’s ability to try new services rapidly and inexpensively or repurpose infrastructure resources for other use.

2: Cost containment — pricing is presented on a utility basis where the user on pays for the portion of actual usage that is demanded during a given period of time. Additionally no IT skills are required of the user to operate the utility infrastructure.

3: Device independence — users should be able to access their service through a web browser regardless of what type of device they are currently using. There are some practical limitations to this but in general should be true.

4: Disperse centralisation — resources are free to be centralised where it is most cost efficient to host them. Additionally the user should not be concerned as to the location of resources, either all in one data centre or in multiple centres because of their abstraction from these details.

5: Location independence — users should have the freedom to connect to their resources wherever a network is available.

6: Multi tenancy — allows the sharing of resource costs across a large pool of users thereby reducing everyone’s overall costs.

7: Peak load capacity — a user is not required to engineer the infrastructure for peaks. This is done for them in the utility service they acquire.

8: Reliability — is enhanced through the use of multiple, redundant and geographically dispersed compute sites.

9: Scalability — performance is constantly monitored and the infrastructure is dynamically scaled in near real time based on user demand.

10: Security — is greatly enhanced to ensure the pool of users that there is adequate fencing of resources and protection of data, providing an auditable trail of accountability.

11: Utilisation optimisation — most IT departments are only able to use 10-20% of their available compute capacity. Because the cloud is shared by a pool of users, utilisation is greatly increased.

Clearing the mist and soaring in the clouds

In the final analysis, the simplest view is the most powerful: a cloud is simply a compute facility that is accessed in a consistent manner, is not impacted by a user’s current location, is available through any of a number of online access methods and is presented to the user much like an electrical utility where you pay for what you use and the infrastructure responds to the pool of user demands.

The benefits that are extended from the cloud service provider to end users, SMBs and enterprise customers are numerous and are worthy of consideration despite the high level of hype around the cloud.

• Because a cloud is a hosted service, users can focus on their core business rather than focusing a significant portion of their attention on IT support functions for the business.
• Technology refresh and feature upgrades are more cost effective and rapidly implemented in the cloud.
• Scalability is greatly increased and there is no need for the end user to pay the infrastructure costs associated with peak demand.
• Application integration is more easily achieved as there is an expectation that the applications will be presented as a web service.
• Finally a new level of collaboration is enabled as the user’s data and services are consolidated into an entity that is by its very nature virtualised and accessible with enhanced security mechanisms and controls.

The net result is an overall reduction in the costs associated with IT needs of a user’s organisation through the use of a shared facility.

For the service provider the benefits are equally compelling. New applications can more rapidly be developed and vetted within the user community. The costs associated with trialling new applications are greatly reduced and the service provider is liberated to try more new services in an attempt to attract new customers and add new application-based revenue streams.

Intimacy with customers is improved as the service provider becomes a partner in business with the end user. New business opportunities more readily present themselves as the customer base shares a common set of services and delivers feedback to the service provider.

As the cloud settles to the ground and becomes a reality it is certain that some fog will appear.

But as this new era of computing begins and the mists of the early definitions and attempts start to clear, new business models will emerge to create a revenue ecosystem from which all can potentially benefit.

Jeff Edlund is chief technologist, futures, in Hewlett-Packard’s communications and media solutions division