Multicloud vs. hybrid cloud management: Top approaches, tools

There is much confusion about what constitutes a hybrid cloud versus a multi-cloud. Because hybrid cloud and multi-cloud are hot terms, it's not unusual for marketing teams to spin whatever they're selling to be one of the two, or sometimes both. 

Hybrid clouds, also called Hybrid IT, combine on-premises infrastructure (private clouds) with public clouds. The term for pairing traditional on-premises systems with public clouds is pragmatic hybrid cloud.

Multi-cloud is a form of hybrid cloud, but it's a specialized term used to connote running in multiple different public cloud environments. It’s usually part of a strategy  to avoid lock-in with a single cloud vendor. 

While most multi-clouds don't leverage a private cloud, some do. But multi-clouds always use more than one public cloud.

IT operations pros needs to understand just what they're dealing with and how to successfully operate these systems. Here's a no-nonsense look at how to manage multi-cloud and/or hybrid cloud, along with discussion about best practices and tools.

Hybrid Cloud Management: Forrester Overview of 41 Providers

Understand the tool patterns

Before you consider how to approach each cloud type for IT operations, it’s important to understand the different patterns each presents:

Hybrid Cloud

Multi-Cloud

Always combines private and public clouds, such as an OpenStack private cloud and AWS.

Always involves two or more public clouds, such as AWS, Azure, and Google.

Data is fixed within a private or public cloud.

Data is fixed within a public cloud provider.

Data can be shared between clouds, and databases can span clouds.

Data can be shared between clouds, and databases can span clouds.

Security approaches and tools are different between public and private clouds.

Security approaches and tools are different between public clouds, with some global security tools moving into place now.

Cloud-native workloads are hard to move.

Cloud-native workloads are hard to move.

Focus on native ops tool support.

Focus on third-party ops tool support.

Focus on native cloud usage and cost management.

Focus on third-party cloud usage and cost management. 

Focus on native performance analytics. 

Focus on third-party performance analytics. 

 

Keep in mind that there are no absolutes in this game. While the table generalizes about hybrid and multi-cloud, there are always outlying types of solutions that may not apply to the above, so make sure you understand your own requirements. 

An ops guide to differences and similarities 

Hybrid clouds lack support for third-party ops tools. The reason is pretty obvious. Since hybrid pairs public and private clouds, there is no need to buy a third-party tool to drive operations management on either side, because the native tools that are provided in each respective cloud, private or public, will always work best. 

While some tools do exist that span private and public clouds, native tools are leveraged almost 95% of the time. 

Multi-cloud is a different story. Considering that the public cloud services, such as storage and compute, are redundant and overlapping, they may be operated using abstraction layers such as cloud management platforms (CMPs) or cloud services brokers (CSBs), or a combination of the two. 

It's the job of the abstraction-layer operations tools to hide the differences from those who operate the multi cloud. All they see is an abstraction of storage, or an abstraction of compute, databases, applications, or any other resource that must be managed. They can manage them using a single interface (single pane of glass) that hides the cloud-native complexities from users.

For example, you don’t need to understand the differences between AWS and Microsoft public clouds in order to manage their respective storage platforms. Each public cloud’s storage systems are represented using a single consistent interface that allows the same operations on each storage system but implements each using whatever respective APIs are needed. 

This does a few things:

  • There is no requirement for the IT Ops team to become expert in all public clouds that are leveraged within the multi-cloud. The team needs to learn the tools, and the tools should translate the ops functions to the native functions of the respective public cloud provider. 
  • The tools automate some operations functions across the clouds, such as automated backup and recovery.
  • There is better cross-cloud governance, because you can track usage of all public cloud services, as well as the aggregate use. For instance, it's a well-known trick to leverage storage from two or more public cloud providers, considering that IT operations rarely cross-references the billing.  
  • A common security layer is possible, which means you can leverage systems such as identity and access management (IAM) across clouds, and use a common directory service. 
  • You're abstracted away from constant changes. Public cloud providers constantly add or remove services. Keeping up is a chore unto itself. These systems remove you from having to deal with constant change. 

While there are no magical tools that solve all problems, the tools that do exist today are worth an investment in time and money. Considering that CloudOps is a new science, it's clear that complexity is the enemy of operations management. Solve the complexity problem now, before it translates into operational failures.

Best practices that won't get you fired

Although cloud ops are just getting started for most enterprises, there is a need to define some emerging best practices that can ensure consistency and better operations than those experienced with traditional systems. 

Here are a few to consider.

Planning is your friend

Most ops teams that make game-changing mistakes do so because they lacked a proactive plan to deal with most issues. You would think that it's during this phase that you pick a boatload of tools, both for hybrid and multi-cloud. The strategic use of tools is certainly a best practice, but it's not a numbers game. Many ops teams over-tool and thus make a complex cloud solution even more complex. Keep in mind that complexity is your enemy and leads to ops errors.

There is no single magic tool

For both hybrid and multi-cloud, the downside is that you'll likely need two or three tools to round out your operations approach. This toolset typically includse CMP, security management, and cost governance, at a minimum. These need to be cross-cloud tools, meaning that they support more than a single cloud provider, including private clouds. In other words, the best practice will include more than a single tool, but don't leverage so many tools that they diminish the value. 

Continuously improve

Most operations management teams hate to admit when they're wrong. To be successful at CloudOps, that needs to change. Operations needs to continuously improve over time, meaning that it constantly reviews what works and what does not. If solutions and technology don't add value, they need to be pushed out and replaced by solutions and technologies that do.

Moving from hybrid to multi-cloud

While many organizations are moving from a single public cloud architecture to multi-cloud, some are moving from hybrid to multi-cloud as well. The reasons vary, but most have discovered that the private cloud portion of the hybrid cloud architecture is not as useful as they anticipated, and so instead opt to leverage more than one public cloud as their target.

The operational impact is positive because the solution has more tools and technology built for it. More native tools, if needed, are also available. Indeed, the reason that most enterprises move to multi-cloud is the availability of a rich set of features and functions in the public cloud. Thus, retooling for a move from hybrid to multi-cloud should be a positive move, whereas moving from multi-cloud to hybrid is the opposite.

Complexity is part of the deal

Keep in mind there are other cloud services that you need to operate, such as databases, security, workloads, and specialized services such as machine learning. While for some of these you can hide their complexity behind operations management tools, you'll have to deal with some natively, using some native public CloudOps tool.

If this sounds complex, it is. Although IT operations continues to remove complexity from CloudOps, the ongoing maturation of cloud computing platforms—as well as the movement of both workloads and data—does not allow enough lag time to eliminate complexity. 

It's almost a whack-a-mole situation. You have to stay alert to keep pace, and it's easy to fall behind. Even with the use of specialized tools and their evolution, it's still going to be a bit of a street fight for IT Ops—at least for the next five years. 

Image: Wikimedia Commons.

[ Upcoming Webinar (Oct. 23): Simplify Discovery and Change Management for Cloud and Container Environments ]