How to architect a hybrid cloud that combines on-premises and public cloud infrastructures. Definitions of hybrid cloud vary, but here’s what it means to Amazon, Microsoft and Google.
A hybrid cloud is a computing platform built from both private and public cloud components. A public cloud is what usually comes to mind when we talk about cloud computing: storage and compute resources offered by a vendor to customers who pay on a metered basis and don’t have to worry about provisioning and managing the underlying infrastructure.
One drawback to using public cloud resources is that they often run in virtualized environments, and customers share hardware and other resources. As an alternative, a customer could set up a private cloud themselves on their own infrastructure, offering the same sort of flexible access to compute resources to internal users.
The customer has much more control in this scenario to ensure security, data privacy, and access to compute resources, but they’re faced with the infrastructure costs and management burdens that caused many to turn to cloud computing in the first place.
A hybrid cloud ideally lets organizations take advantage of the benefits of both public and private clouds while mitigating the respective disadvantages. For instance, a company may set up a private cloud infrastructure in its own data center, but offload some compute cycles to a public cloud rather than pay to beef up their in-house server hardware to handle occasional high computing loads.
Another organization might rely on a public cloud vendor for most of its infrastructure needs, but use on-premise servers to store and deal with sensitive customer information in order to conform with strict data security and privacy laws.
Simply using both private and public clouds isn’t enough to make a hybrid cloud platform, though. The National Institutes for Standards in Technology (NIST) defines a hybrid cloud as follows:
(Hybrid) cloud infrastructure is a composition of two or more distinct cloud infrastructures (private, community, or public) that remain unique entities, but are bound together by standardized or proprietary technology that enables data and application portability.
That “bound together” part is important: hybrid cloud computing means that, from the perspective of the user, the public and a private cloud resources appear as much as possible to be a single environment, or at the least have the ability to be managed by the same set of tools.
Multicloud architecture is not the same thing as a hybrid cloud, despite similar names. A company pursuing a multi-cloud strategy is using multiple public clouds—both Amazon AWS and Microsoft Azure, for instance. These multiple clouds may end up being part of a hybrid cloud architecture, but without the private cloud component, it’s not a hybrid cloud.
How does a hybrid cloud work?
The fundamental problem that a hybrid cloud solution needs to solve is how to connect its private and public cloud components as seamlessly as possible, while still giving adminstrators granular control over where data resides and compute cycles execute.
A company that wants to offload some of its compute cycles from its datacenter to the public cloud is probably okay with automated management tools simply moving some jobs to the public cloud on an as-needed basis.
But an organization concerned about complying with privacy laws will need the ability to ensure that certain data and compute jobs only reside on-premises. Hybrid cloud relies on the right management tools to make all this possible.
One important advancement over the past decade has been the move to container-based microservices managed by platforms like Docker and Kubernetes. An orchestration platform like Kubernetes can do much of the work of distributing workloads across the components of a hybrid cloud architecture, making a hybrid cloud easier to use and more attractive to companies intrigued by its advantages.
What are the types of hybrid cloud architectures?
There are several architectural approaches to hybrid cloud.
Vendor-native hybrid cloud. An organization building a hybrid cloud architecture may begin by choosing a public cloud infrastructure-as-a-service (IaaS) offering. They can then provision in-house resources to align with their choice of public cloud, essentially extending that public cloud into their own data center to ensure maximum compatibility. This is known as vendor-native hybrid cloud, and all the all the major IaaS cloud vendors have made this easier to do.
For the on-premises extension to the public cloud, companies including VMware, Red Hat, Hewlett Packard Enterprise, Cisco, Dell, and IBM enable customers to manage public cloud resources. Meanwhile, public IaaS providers are building connections from their public cloud into their customers’ data center.
Hybrid cloud management software. A plethora of startups and established infrastructure-management vendors have developed software that allows users to centrally manage both on-premises and public cloud infrastructure and applications. From a single console, virtual machines, storage, databases, and other resources can be spun up and down, regardless of whether they’re in a company data center or the public cloud.
Forrester Research analyst Lauren Nelson says hybrid cloud management was much-heralded in the early days of cloud computing, but it did not develop as robustly as some expected. Many IT pros have found it sufficient to manage public cloud and on-premises resources using Application Programming Interfaces (APIs) and tools native to those platforms that provide deeper functionality than overlay management software.
As a result, many of these software systems have evolved to focus on specific tasks, such as cost control, application performance monitoring, or allocation of resources.
Platform-as-a-service (PaaS). Many organizations utilize platform as a service (PaaS), which enables developers to write custom applications without provisioning the underlying infrastructure.
Examples of PaaS include Pivotal Cloud Foundry, Red Hat OpenShift, IBM Bluemix, and Apprenda. Most major PaaS software can run on premises, hosted in a private environment or natively in the major IaaS public cloud. The PaaS automatically configures infrastructure resources across these environments, making them a platform for hybrid cloud.
Hybrid cloud benefits
Hybrid cloud is all about flexibility: you can take advantage of either a private or public cloud, depending on your needs in any given situation. That means:
- You can choose the platform that provides the optimal application performance in any given scenario, based on compute needs and network latency. For instance, you could run a SaaS office suite in-house to provide low latency for users, but shift monthly big data batch processing to the public cloud for low-cost number crunching.
- You can apply granular application and data governance, deciding whether to make use of local compute and storage resources or the public cloud, based on regulations, security, and other considerations.
- You can make the most cost-effective use of resources, with a balance of capex and opex spending. For instance, by shifting occasional large compute jobs to a public cloud, you can avoid the need to spend to boost your in-house systems.
- If a total shift to a public cloud is your ultimate goal but your on-prem data center still has useful life left in it, a hybrid cloud can be part of an incremental migration.
Hybrid cloud challenges
There are also challenges to adopting a hybrid cloud that must considered before any migration:
- Private cloud provisioning and maintenance: The most tempting promise of cloud computing is that it abstracts the user away from the underlying architecture, relieving you of the burden of maintaining and administering server hardware. With a private or hybrid cloud infrastructure, however, you still need in-house sysadmins and other IT staff to care for your datacenter.
- Data handling and network security: Traffic between the private and public portions of your hybrid cloud is less secure than internal traffic within each component.
- Access management: Your infrastructure must try to make access as seamless as possible across your hybrid cloud platform, allowing users and processes access to the resources they require, without compromising security.
- Skills requirements: If your system and network admins are primarily familiar with in-house datacenters, they need to get up to speed on what the public cloud requires—and vice versa.
Hybrid cloud platforms
Despite an earlier reluctance, major IaaS vendors have in recent years have embraced integrating their public cloud infrastructure with customers’ on-premises resources. Vendors are creating tools that work across these environments and they’re partnering with companies that have strong ties in enterprise data centers.
Amazon Web Services
AWS, viewed by most as the public IaaS cloud leader, was initially reticent to market itself as a hybrid cloud company, instead focusing on its public cloud capabilities. That changed in a big way when the company announced major partnership with VMware in 2016. Subsequent rollouts include Outposts, a hybrid cloud service announced in 2019, and EKS Anywhere and ECS Anywhere, which help run containerized workloads seamlessly across a hybrid cloud architecture.
Microsoft Azure
Microsoft has by far been the most vocal in marketing itself as a hybrid cloud company, which is unsurprising given its incumbent status in most IT shops already. Microsoft’s primary hybrid cloud platform is Azure Stack, a converged infrastructure hardware distributed by Dell EMC, Cisco, Lenovo, and HPE.
It’s meant to run just like the Azure public cloud, but it sits on customers’ premises. Microsoft also allows many of its cloud-based software management tools to be extend on-premises, as well as its broad range of SaaS applications, including Office 365 and Outlook. The company has also rolled out a hybrid cloud server management platform known as Azure Arc.
Google Cloud Platform
Like AWS, Google focused its early marketing efforts on public cloud, but the company is now providing hybrid cloud offerings as well. The open-source Anthos platform, rolled out in 2019, is the foundation of these efforts. Anthos allows customers to deploy and manage Kubernetes workloads across a variety of environments, and underpins a number of distributed cloud options the company unveiled in late 2021.
The future of hybrid cloud
One thing that seems almost certain as more companies push into digital transformation is that the future of the cloud is hybrid—and indeed, that the distinction between public and private clouds will get blurrier and blurrier.
With a hybrid cloud architecture, you can extend your public cloud provider’s services to your local infrastructure; you can also rent private physical or virtual servers from colocation facilities or even public cloud providers themselves, and hybrid cloud technologies knit all these together.
The hybrid cloud revolution is allowing organizations to harness together all kinds of infrastructure to create heterogenous architectures that best meet their business, technology and financial needs.