Cloud deployment models compared

In this first of five articles on planning your journey to the cloud, we'll take a look at the four main cloud deployment models and what they might mean for different types of businesses.

The US National Institute of Standards and Technology (NIST) came up with a dry-but-effective definition for cloud computing back in 2011. Some parts of it are outdated, but it covers essentials that remain relevant to this day. The seven-pager (which contains five pages of disclaimers and bureaucratic digressions and two whole pages of actual content) outlines the different service models, deployment models, and essential characteristics of the cloud.

The essential characteristics—on-demand self-service, broad network access, resource pooling, rapid elasticity, and measured service—seem a bit obvious in this day and age, but cloud deployment and service models are some of the most important concepts to understand if you're considering migrating IT workloads from an on-premises environment to the cloud.

What are cloud deployment models?

The saying that the cloud is just someone else's computer may well be a cliché by now, but clichés exist for a reason. A cloud deployment model, simply put, refers to the nature of that "someone else's computer": who owns it? How is it managed? What workloads does it house?

The NIST defines four cloud deployment models:

Let's take a look at each one, its pros and cons, and why it may or may not be a good fit for your organization's cloud-computing needs.

Public cloud

The public cloud deployment model is what typically comes to mind when we think about cloud computing. In this model, a public cloud provider, such as Amazon Web Services (AWS) or Microsoft Azure, owns all the infrastructure required to run an organization's IT workloads. Keeping with the "someone else's computer" trope, hosting workloads in the public cloud equates to renting as much computing power and storage space as you need from someone else's giant, infinitely powered PC.

In a recent RightScale survey of 786 technical professionals from both enterprise-level and small-to-medium businesses, 91% of respondents reported hosting at least some of their workloads (applications, document storage, etc.) in the public cloud.

Smaller organizations especially favor the public cloud over other deployment models: 44% of SMB respondents reported using public cloud exclusively, from either a single vendor or multiple.

This makes sense: hosting your workloads in the public cloud means you don't need to worry about purchasing, supporting and maintaining any costly infrastructure; instead, a giant corporation like Amazon or Microsoft takes care of it all for you while your team works on actually building things. The absence of an up-front cost removes a significant barrier to entry for smaller organizations with ambitious development goals. In the cloud, you only pay for what you need.

Nearly half of SMBs host all cloud resources in public clouds.

This turnkey approach applies to security and reliability as well. Rather than hosting operations-critical or otherwise sensitive assets on physical in-house servers (as you would in a traditional on-prem environment), you're entrusting them to a highly secure data center with more built-in redundancies than you could ever fathom. To give you an idea, Microsoft's SLA guarantees at least 99.9% availability on the vast majority of its paid services, and up to 99.99% for some critical services such as DDoS protection.

That's more than enough for a large chunk of businesses. But even 99.99% availability is insufficient for some types of organizations, especially in highly regulated fields like healthcare and government—hence the remaining three cloud deployment models.

Private cloud

As far as the end-user experience goes, there isn't much difference between public and private clouds; all those essential characteristics of cloud computing (resource pooling, on-demand self-service) we briefly mentioned earlier apply to both.

Infrastructure-wise, a private cloud is essentially a smaller-scale replica of the datacenters owned by public cloud vendors. Private cloud deployments are much more expensive than their public cloud counterparts, but they offer a variety of benefits that make the greater up-front costs worthwhile.

Businesses rarely go the public cloud-only route: in the aforementioned survey, only 3% of respondents used this model exclusively to host their cloud workloads. However, 75% reported using private cloud to some extent, hosting some workloads privately and entrusting others to public cloud providers. This is what's referred to as a hybrid cloud.

Three-quarters of businesses host some workloads in a private cloud.

Hybrid cloud

As its name suggests, the hybrid cloud deployment model refers to a combination of private and public clouds used to host an organization's workloads.

For many, hybrid represents the best of both worlds, which explains why a vast majority of businesses—69%, according to one survey—choose this approach. (This isn't the same thing as a multi-cloud strategy, which refers to relying on multiple cloud vendors in order to maximize availability and cost).

69% of businesses prefer a hybrid approach that balances both public and private cloud use

Depending on your needs, there are some potential downsides to a hybrid cloud deployment. Setting up a private cloud usually requires up-front investment, negating some of the cost-saving benefits of going full public. It also involves managing your own infrastructure to some extent. However, if you have a lot of legacy to support yet still wish to benefit from the public cloud for your other workloads, for instance, hybrid is often the best-case scenario.

Community cloud

For the sake of exhaustivity, I've included all four NIST-defined cloud deployment models in this article, but this last one is by far the least applicable to most situations. It is, however, an interesting approach.

Basically, the community cloud is run like a cooperative. In a hybrid cloud deployment, some of your workloads will be hosted in the public cloud, while others are kept in privately owned and maintained datacenters. The community cloud, on the other hand, combines aspects of the public and private cloud in a single, unified environment.

Per Gartner's always-handy IT glossary, "Community cloud computing refers to a shared cloud computing service environment that is targeted to a limited set of organizations or employees (such as banks or heads of trading firms)." These groups of organizations generally share a set of interests and requirements (usually related to security, compliance and/or jurisdiction) that can't be met by any of the public cloud vendors. Instead of each building their own private cloud environment, they pool their resources in order to build a single infrastructure able to support the needs of each stakeholder. Use cases for community cloud deployments are limited, but this model comes with highly advantageous cost savings for those to which it applies.

Hybrid vs. public vs. private cloud: which one should I choose?

The answer, of course, is "it depends". While hybrid is a good option for most businesses, some may have requirements that would be better met by a public- or private-only approach.

  • If cost is the main decision factor and you aren't in a heavily regulated industry, definitely consider hosting your workloads in either a single or multiple public clouds.
  • If you have particularly stringent security and compliance requirements (due to industry regulations, for example), or if your organization has a bunch of legacy to support, you'll likely need to host at least some workloads in a private cloud. However, it very rarely makes sense to host 100% of your workloads privately.
  • If you own, say, a bank, and are contemplating a cloud migration project, consider banding together with a few fellow financial institutions to build a mutually beneficial community cloud.
  • Otherwise, hybrid deployments provide the most flexibility and control while still allowing you to benefit from the public cloud's cost-saving characteristics.

Next up: cloud service models, explained

Check out the next instalment of our cloud computing fundamentals series for a look at the main cloud service models available: software as a service (SaaS), infrastructure as a service (IaaS), platform as a service (PaaS), and function as a service (FaaS).

If you're interested in learning more about the cloud and its potential for businesses of all sizes, be sure to download your copy of our new book, Modern Business Powered by Microsoft Azure, by Azure MVP Jussi Roine in collaboration with the team here at ShareGate.


You might also like