A Stacked Guide To Cloud Infrastructure

Clouds are nebulous masses of vapor that are forever changing shape. At least, the ones in the sky can be defined in this way. Talk to some non-technical business managers, however, and you find they have a similar conception regarding clouds of the computing variety. Let’s take a closer look at their digital counterparts.

A Brief History Of Cloud Infrastructure Time

Although the concept of “the cloud” has been with us for well over a decade, the initial hype from all corners of the IT industry sewed seeds of confusion that continue to grow to this day. Businesses labelled all manner of products and services as “cloud-enabled”, “cloud-based” or “cloud-ready” in a bid to jump on what they saw as a lucrative bandwagon.

Was the cloud just another word for the internet? For virtualization? For what used to be called ASP (delivering applications over a network as a service)? Depending with whom you spoke, it could be any and all of these things plus a whole lot more.

The truth was — like many broad technology concepts — in the early years different definitions and models of “the cloud” and “cloud computing” were jostling for position. The term had no fixed meaning. But as the industry has grown and matured, all these different ideas have coalesced into a standard, shared understanding of what “the cloud” is, and of the various elements that constitute cloud infrastructure and architecture.

So, for anyone still suffering from the hangover of early-years cloud confusion — or who wants to be able to explain the concept simply to non-technical managers and budget-holders — we’ve set out below an at-a-glance outline of cloud infrastructure.

The Building Blocks

• Like many tech areas before it, cloud infrastructure is best pictured as a ‘stack’ — where ascending layers build on the ones below, each representing a further level of abstraction from the base hardware and software on which they depend. You might call them the cumulonimbus stacks if you had a mind.
• At the base level sit the fundamental hardware components required to build any cloud — servers, storage and networking.
• An abstraction layer known as a hypervisor sits above the base physical layer. This virtualizes hardware resources so that users can set up virtual machines (VMs) which — although they are defined and set up entirely by software — to all intents and purposes operate and appear to the user as distinct, separate computers.
• However, because these VMs are not tied to any particular hardware resources, they can be scaled up and down as necessary and can make use of hardware resources irrespective of where these are located physically.

• The hypervisor can either be based directly on hardware or run on top of other software. They are known respectively as ‘bare metal’ (type 1) hypervisors, which sit directly on the hardware, and ‘hosted’ (type 2), and run on top of an existing operating system. Because the latter are more abstracted, they tend to be slower, but on the plus side they are able to support a greater variety of systems.
• Popular bare-metal hypervisors include VMware ESX,Microsoft Hyper-V and Citrix Systems XenServer.
• Popular hosted hypervisors includeVMware Workstation, Microsoft Virtual PC, Parallels Desktop and Oracle VM VirtualBox.
• Your choice will depend on your current IT infrastructure, your particular business requirements as well as weighing up factors such as processor overhead, amount of memory and virtual processors supported.
• Other essential building blocks of cloud infrastructure include load balancers, network switches and firewalls. Cloud service providers, in particular, must ensure clients’ VMs can share hardware resources without leaking data or hogging resources.

Stacking Up Cloud Services: IaaS, PaaS and SaaS

• With hardware resources and hypervisor in place, we can build up the cloud stack, starting at the bottom with infrastructure as a service (IaaS). This refers to the on-demand provision of virtual machines with associated storage, computing power, memory and bandwidth over a dedicated network or the Internet.
• The next layer up is platform as a service (PaaS), which also provides the operating software and tools necessary to deploy applications without having to worry about configuring virtual machines from scratch.
• At the top of the stack is software as a service (SaaS) where full applications and/or databases are provided on demand over the network as a service.
• You might also hear cloud infrastructure providers talking about other ‘as a service’ models — such as DaaS (desktop or data as a service), BDaaS (big data as a service), SECaaS (security as a service) and more — but these can generally be considered a subset of either SaaS or PaaS rather than any fundamental addition to the stack.

Cloud Hosting, Ownership And Payment Models

• Virtualized resources at any level of the stack can either be hosted by the organization’s IT department or by third-party service providers.
• Where resources are hosted by third parties, they are today generally paid for on a usage basis, either weekly, monthly or longer.
• As cloud data center technologies mature further, it is expected that more providers will introduce true metered ‘utility-style’ cloud service models — where users pay for precisely what they use.
• Different providers offer different plans and levels of service, which may include options to have dedicated hardware for the organization’s VMs rather than share the provider’s hardware with any of its other clients.
• Cloud infrastructure owned by the consuming organization (or rented on an exclusive basis from a third party), is known as private cloud while cloud infrastructure owned by a third-party provider and shared among its clients is known as public cloud. Where an organization uses a mixture of the two, they are said to use hybrid cloud.

Of course, there’s a host of other cloud computing terminology you might come across, but if you can master the basic concepts outlined above, you have enough knowledge to end cloud confusion and call out any unnecessary hype and nonsense.

Like what you read? Sign up for our weekly newsletter at 100TB.com.

Originally published at blog.100tb.com.