One of the big themes at this year’s AWS Re:Invent conference was that fact that AWS is moving fast to support hybrid-cloud computing. Of course, AWS has no aspirations to purchase or build a private cloud (not counting the CIA cloud). Instead, it wants to support more hybrid options using partners and creative technology approaches such as virtual private clouds (VPCs).
It’s no secret that most enterprises are moving toward hybrid and multicloud computing. In some cases there is a clear business case. At other times it just seems like hybrid is the approach that gives an enterprise the most options and the most control.
The problem with the hybrid-cloud approach to cloud computing is that many enterprises see it as a way to avoid a commitment to a specific technology solution or architecture. Hybrid clouds are made up of private and public clouds. In many instances, they are made up of traditional systems, devops, public PaaS, and many brands of public clouds (e.g., AWS, Google, Microsoft, etc.). Thus, hybrid becomes more of a platform for building a solution than a solution itself.
Enterprises move to hybrid and multiclouds for a variety of reasons:
- Single-cloud solutions typically don’t provide the breadth and depth of functionality that enterprises require for all of their cloud-computing solutions, so they mix and match public and private options.
- The rise of cloud management platforms (CMPs) gives enterprises a single interface to help provision, manage, and scale complex cloud environments.
- Companies that want to move applications into public clouds need a range of services, including different databases, middleware, development, and compute services, and this drives the use of multiple cloud computing platforms.
- The growing use of platforms, infrastructure, and software in the cloud results in multiple forms of clouds.
- The ability to maintain control of applications and data as needed to address specific security and compliance requirements.
While such a move is attractive, the best approach to hybrid cloud is to understand what solutions an enterprise can likely provide out of this architecture. Many are looking to set up an all-you-can-eat cloud buffet, but there are some choices to be made in terms of what’s included in a hybrid or multicloud architecture to maximize the business benefit.
Most enterprises have very complex requirements for applications that may span more than a single platform. In the case of cloud computing, the use of either private clouds or traditional systems, when paired with public clouds, provides a few significant advantages:
- The ability to move workloads from owned hardware to public-cloud services as the requirements of the workloads change.
- The ability to split or distribute workloads between on-premise and remote public-cloud services, thus providing the ability to scale out, or, “cloud burst” as needed to accommodate changing workloads.
- The ability to mix and match best-of-breed cloud services, such as leveraging a cloud database service from a public-cloud provider, with an application that may be on-premises.
As with any complex approach to computing, there are tradeoffs to consider. In the case of hybrid-cloud computing, the ability to track the consumption and utilization of resources becomes much more difficult, as does cloud service management, governance, and security. The more complex your hybrid or multicloud architecture, the more it will likely cost and the more risk you’ll add to mix.
Most enterprises will or do leverage a hybrid-cloud model, and management needs to keep their eyes open around the risk and costs it is likely to encounter. This means creating and managing the right logical approach to a hybrid-cloud architecture and solution, and making sure that the right resources are leveraged for the right reasons.
Picking a path
As you can see from Figure 1, there are a few core component categories that should be considered, including a private cloud and traditional systems, that make up those things that are on premises. On the remote side, we have public IaaS (e.g., AWS EC2 and S3), PaaS (e.g., Google App Engine), SaaS (e.g., Salesforce.com), and other cloud services. Other cloud services are typically tactical services, such as cloud APIs to financial data, weather data, etc., used to support certain applications. And governance, management, and security become systemic to the entire configuration, and span both public and private aspects of this architecture.
Figure 1: Hybrid clouds are made up of many architectural components.
We’re using this model to depict some of the logical hybrid-cloud architecture components that should be dealt with or considered. The trick is to pick a path through this architecture that best addresses the business and technology requirements of your organization.
As seen in Figure 2, enterprises need to pick an area within the hybrid cloud to focus their use. This example focuses on the use of traditional systems, as well as IaaS public clouds, some private cloud, and the inclusion of one or many SaaS clouds. The idea is for companies to place specific bets on which hybrid cloud subsystems to build that map back to their requirements.
Figure 2: Those building a cloud strategy need to pick a specific path through the options before they begin onboarding hybrid cloud components/services.
In this case, it would not make sense to spend a lot of time and money onboarding a public PaaS, or even spend a great deal of money and time on the private cloud. This would be architecture more about traditional systems that are able to leverage some public-cloud assets as needed. Although many of those who define “hybrid cloud” view it as a paired public and private cloud, a private cloud is not always a requirement. Those that don’t or won’t need it shouldn’t include it.
Those moving toward hybrid or multiclouds need to do the following:
- Define the business objectives of IT, and the ability for a hybrid cloud solution to add to meaning to those objectives.
- Define the solution patterns needed for both short and long term. What hybrid cloud components will likely be needed? This includes type (e.g., public IaaS) and purpose (e.g., mass raw-data storage).
- Review the ability of the hybrid cloud component to provide the right technical solution, such as performance and features.
- Define the role and mechanisms of management, security, and governance.
- Define a phased implementation plan, including timelines and business value gained throughout the implementation. And define budgets and risks to not obtaining business value.
- Execute around short projects — typically less than four months. For example, onboarding of a public IaaS provider, integration with traditional systems, installation and deployment of a private cloud, etc..
While the use of hybrid and multicloud approaches are attractive for enterprises as they move toward cloud computing, there are a huge number of mistakes that can be made that can cause the overall strategy and effort to fail. The largest issue is trying to address any potential requirements without first understanding the core requirements and objectives of the hybrid cloud. While it’s nice to keep options open, the likely outcome is no delivered value.
Enterprises must be diligent to understand their requirements and how to address them. Hybrid clouds should be built and designed to meet specific needs, not general solutions. That’s the only way to make them pay off.