Hands holding a cloud

5 essentials for managing hybrid cloud

If you use cloud computing, chances are good that you leverage some sort of hybrid cloud as part of your environment. Until recently, this usually meant pairing a private and public cloud, such as OpenStack and Amazon Web Services. Today, the term “hybrid” typically means legacy, or traditional, systems paired with one or more public clouds, a single private cloud that interfaces with two or more public clouds, or any combination thereof.

Things get complex, quickly. As enterprises attempt to figure out the best approaches to security, governance, and management, they're finding that no single approach or tool can solve all problems. So, if an enterprise IT shop has a hybrid cloud or, more likely, more than one hybrid cloud, what should it do about management?

The first step is to understand the essentials. When defining them, it’s not about just picking the tools and technology you'll use. Most IT operations managers make the mistake of focusing on tools that may make hybrid cloud management easier, rather than understanding their own requirements, which means they get both the approach and the tool selection wrong. You need to understand the security, data, governance, and end-user dynamics that affect how you approach hybrid cloud management. 

Continuous Testing: A Practical Guide

Here are five concepts that you need to understand before moving to any hybrid cloud management platform.

1. Understand what’s being managed

While this seems like the single most important thing to know, many of those who define a hybrid cloud management strategy fail to understand the profiles of the workloads that will run on public and private cloud(s). You need to understand what the applications do, including how they interact with the end users, manage data, how they handle networking, security patterns, performance, etc..

Specific things need to be understood, as follows: 

  • Who owns the workload within the organization? Who needs to be contacted when things go south?
  • What do the workloads do for the business, per their criticality to the business? This goes to how many resources you spend on managing the workloads on the hybrid cloud, which need to align with the value they bring to the business.
  • When do the workloads run? Some run continuously, while others may run during the same hour in the day. Again, this goes to how you approach the management of workloads within the hybrid cloud.
  • Where do the workloads run? On the public cloud, private cloud, or in both places?
  • Why were the decisions made about where to run the workload? And when they may need to be re-evaluated?
     

2. Understand security and governance

These days, security and governance are a requirement, whether a mandate from your customers (see “SLAs” below) or from your senior management. This means you need to proactively manage security to make it work. You can also leverage new mechanisms such as IAM (identity and access management), which allow assigning of identities to data, people, devices, and servers, to configure who can access what, and when. Finally, information needs to be encrypted at-rest in some cases, and in-flight in others.

Core to this part of hybrid cloud management is how you deal with a few issues:

  • Security and performance. If the needs of the workload are that information be encrypted at-rest (on the storage systems in the private or public cloud), or in-flight (moving over the network), that may result in the risk of lower overall performance. That needs to be understood and managed, including the use of performance monitoring tools.
  • Policy management. Governance requires that policies are written and enforced, and this enforcement needs to be understood by those who are managing the hybrid cloud so that they do not conflict or otherwise get in the way of operations.
     

3. Build a “single pane of glass”

Those who manage hybrid cloud manage complexity, because the private and public clouds all come with their own native APIs and resources. Indeed, they all manage storage, networking, provisioning, and security differently. Thus, you can either learn all of the native interfaces for all private and public clouds, or you can instead build a single pane of glass that abstracts you away from that complexity.

There are tools that can manage the cloud services using a single interface to translate what something means on one cloud versus another cloud. For instance, you need to monitor performance on Google Cloud Platform, and OpenStack private cloud, and Amazon Web Services. All provide different approaches and interfaces to manage performance, and the single pane of glass interface deals with the differences on your behalf, translating what’s important to those who manage the hybrid cloud in and between the different clouds that are under management.

4. Understand the SLAs

SLAs, or Service Level Agreements, are a contract with the end users stating that you, the hybrid cloud manager, and the cloud providers themselves, will supply a specific level of service, else there will be penalties. While you can certainly pass the buck to the public cloud provider in living up to their own SLAs, the hybrid cloud itself is your baby, and thus you’ll be held responsible if the system misses the limits outlined in the SLAs you’ve agreed to.

At a high level, what’s defined in the SLA needs to be defined in the management layer as well.  It’s not just about providing a baseline of good performance to the end users, but it’s about providing performance that meets specific expectations.  For instance, the ability to provide a sub-second response to the sales person leveraging the inventory application that exists within the hybrid cloud.

When it comes to hybrid cloud management, SLAs are not legal tools. But they are a way to define user and business expectations. Thus, it’s easy to leverage these expectations to define the service expectations that need to be managed by the hybrid cloud management layer, and the hybrid cloud managers.  Use them as guidelines.

5. Understand the tools

Many charged with hybrid cloud management often focus too much on the management tools that are available. These tools cover areas such as API management, resource management, cloud management platforms, performance management, DevOps management, security management, network management, native platform management, etc..

There can be as many as a dozen or so tools that you’ll need if you’re going to manage a hybrid cloud effectively.  Picking tools is a matter of understanding what we’ve defined above as the requirements patterns, and then figuring out the solutions patterns that will match. For instance, if your requirement is that you encrypt all data at-rest or in-flight, and you need to manage both the encryption and the performance, then you would look for a tool or tools that provide both types of solutions.

It’s then a matter of working through the requirements, defining the common patterns, and then matching up the tools with the pattern. Sometimes you won’t find all of the tools that you need for all of the patterns. In those cases, you may consider custom tools.  Or, perhaps a hybrid cloud is not in the cards after all, if management of those workloads on those platforms can’t be done using automation through management tools.  In other words, in some cases, it’s OK to leave a workload off a hybrid cloud if they can’t be managed properly or at a reasonable cost.

Hybrid cloud growth and adoption patterns

Markets and Markets reports, “The hybrid cloud market is estimated to grow from USD $33.28 Billion in 2016 to USD $91.74 Billion by 2021, at a Compound Annual Growth Rate (CAGR) of 22.5% during the forecast period.”

Of course, the issue with hybrid cloud is that the analysts and technology providers are tossing many things into that category. For our purposes, we can call a “hybrid cloud” any collection of systems where you have at least one private cloud and one public cloud working together to support systems for IT.

However, managers also need to understand the emerging notion of the “pragmatic hybrid cloud.” This is a traditional set of systems, typically running in a data center, paired with at least one public cloud. This configuration is growing in popularity, as many enterprises continue the migration to cloud, but they do so without leveraging a private cloud. Instead, they pair their traditional systems with public cloud-based systems.

The emerging patterns are ones of increasing complexity. The right way to manage these combinations is not straightforward, because management needs to deal with the increasing complexity as well. This presents a new problem area, because one of the core management reasons for hybrid cloud adoption is to hide complexity behind an abstraction layer. In fact, instead of avoiding the complexities associated with hybrid cloud, managers need to better understand the native features and capabilities of the public or private clouds that make up the hybrid cloud.

Cloud management: Still more art than science

Hybrid cloud management is still more of an art than a science, considering that we’ve not been at it for very long. The complexity will likely increase over time, and that will mean that our approaches to hybrid cloud management need to evolve as well.

For now, it’s a good time to understand just what you’re dealing with in terms of technology. If you're charged with management of these platforms, you have your work cut out. 

Continuous Testing: A Practical Guide
Topics: IT Ops