The rise of hybrid cloud and multi-cloud, as well as hybrid IT, has led to the creation of a whole sub-science of tools, approaches, and techniques to manage complex and distributed cloud deployments.
While some of these are sophisticated enough to leverage deep abstraction layers and simplify the management of a hybrid cloud, most are ad hoc and chaotic. The result: Many organizations don't know where to start, and don't understand when they are making the wrong choices.
But there's a path to complex, distributed cloud management (CDCM) that has yet to be mapped. You start by understanding your own requirements, learn about CDCM at its most primitive level, and then determine the best approach for both management and tooling.
Here's what's new, and what's better, with CDCM—in other words, what's not widely well known or understood, and the secrets to success.
(If you need guidance on the fundamentals before proceeding, check out "5 essentials for managing hybrid cloud," which covers how to understand what's being managed, security and governance, building a "single pane of glass," SLAs, and the tools available to you.)
Cloud complexity: Where are the tools?
I'm defining "hybrid cloud" using the classic NIST definition, but many vendors have taken liberties with the term, and in some cases their interpretation has little to do with a true cloud. To be clear, a hybrid cloud is a paired private and public cloud, the core components of a hybrid architecture.
Multi-cloud is a bit more complex. While you could argue that a hybrid cloud is also a multi-cloud, the industry has defined multi-cloud as two or more public clouds within an architecture. It may or may not include private clouds.
These days, most cloud architectures are complex and far-reaching. This includes traditional systems paired with public clouds, IoT-based private clouds paired with public clouds, multiple private clouds paired with single or plural public clouds, and so on. Indeed, it's any cloud architecture and deployment that extends to more than one public or private cloud.
Those charged with cloud management are feeling the stress, given the chaotic and complex nature of the tools. Many practitioners view tools and best practices to be two years behind the emerging complex hybrid cloud architecture that organizations are using today.
Here's what a cloud architect, developer, or administrator can do to manage their complex cloud deployments in a repeatable and reliable way.
Understand your cloud and your requirements
Most of those who own a complicated, distributed cloud do so without a good understanding of what they are managing. This leads to the misapplication of tools and approaches, and ends with shops too often using management approaches and tooling that could derail an operational complex cloud deployment.
The best way to manage a complex cloud is to understand its nature. This includes the architecture that encompasses which private clouds are leveraged and which platforms they externalize. Moreover, you must gain the same understanding of any public clouds that are leveraged as part of your hybrid cloud or multi-cloud deployment(s).
The focus should be on requirements, or what you plan to do with your hybrid cloud or multi-cloud. For most enterprises, this means you need to understand the cloud's business use and which applications are paired with databases that will either be distributed between the private and public cloud instances or run in one tier or another (on either a private or public cloud).
Or perhaps you're running systems inter-cloud, such as spreading applications and data across Amazon Web Services (AWS) and Azure, for example.
In this set of requirements, you need to understand:
- Which databases you leverage, such as MySQL or Oracle
- Which programming language you use, such as Python or Java
- Which platform you run on, such as Linux or Windows NT
You must repeat the same set of questions for all private and public clouds that are a part of your architecture.
Understand your management interfaces
Next, you need to define the management interfaces into those technologies, which include, at a minimum, the database, application, and platform. You get bonus points for understanding and defining the interfaces into your security systems, governance systems, and cost accounting systems as well.
Interfaces include API sets for accessing features of those subsystems, such as monitoring and managing processor and I/O saturation points, and the ability to spin up additional resources, and spin them down as needed.
The output of this process varies from one management solution to another.
Management technology available today
A common approach to complex distributed cloud management, or other complex and distributed architectures for that matter, is to toss as many tools at the problem as you can. But this leads to more complexity, cost, and, in some cases, epic failure.
The correct selection of cloud management platform tools, or CMPs, as well as cloud service broker tools, or CSBs, as well as AIOps and other emerging ops tools, will ensure survival or guarantee failure, depending upon what tools you select.
There is enough written about CSBs and CMPs that there's no need to go into details here. If you're not up to date on this technology, get that knowledge before proceeding. Once you have the basic requirements in place, it's pretty easy to map those requirements to the correct tool set.
These tools typically provide a layer of abstraction and automation between the humans who manage the complex cloud deployments, the many clouds themselves, and, most importantly, the different components, including application, data, platform, security, governance, and so on.
Moreover, the right tools have the ability to automate the things that can be automated. Examples include shutting down or restarting server instances that produce I/O errors or that are blocking access from specific IP addresses that appear to be launching a distributed denial-of-service attack.
However, you need to do a fair amount of research, testing, and operational trials to find the correct tool (or tools) you'll need for your cloud-hosted applications. There are about 40 viable tools on the market today. Getting that list down to the final five will take you a month or two. From there, plan on another month or two to test the tools before making your final pick.
It's an easy process to follow. However, success depends in large part on your ability to define your requirements up front.
Create a plan that will ensure success the first time
Now that you have defined the requirements and tools, based on the specific needs of your complex, distributed cloud, focus on the larger-scale planning and budgeting that need to occur next.
These plans differ, depending upon the type and complexity of the hybrid cloud and/or multi-clouds. However, they typically include:
- A high-level solution, including core requirements, objectives, SLAs, and other macro requirements
- A low-level solution, which defines all the APIs, tools, approaches, and other tactical solutions in detail
- The planned budgets for all approaches, technologies, etc.; this includes humans needed over time, technology deployed over time, cost of risk, and ongoing improvements
While this might sound simple, cloud management pros are typically not good at writing things down, nor are they good at understanding costs. I recommend creating a center of business management to do this kind of planning centrally, as opposed to pursuing ad-hoc efforts for each project instance.
Track the key trends
Finally, consider what's coming next. Complex, distributed cloud management solutions, best practices, and approaches are continuously changing. You need to continuously improve, so keep up with the trends. Look for new ideas, new approaches, and new tools that will provide value.
A plan for continuous improvement is a fairly new concept to IT, considering that IT organizations have grown accustomed to releasing 1.0 1.1, 2.0, etc., and have thus found the waterfall way of doing application development—even for complex, distributed cloud solutions—more comforting. However, the progress of technology means that you need to look for better best practices and technologies.
New technology adoption also drives this need, including serverless technology that runs on public clouds such as AWS and Microsoft Azure, as well as container technology that can run on private or public clouds.
There are two macro patterns to consider:
- The evolution of technology that enhances your ability to build solutions on a hybrid cloud, such as serverless and containers
- The evolution of cloud management tooling, such as new CMP, AIOps, or CSB products
Many organizations don't know where to start, or they make the wrong choices when they approach complex cloud architectures. However, the single most important takeaway is that as successful cloud management pro you should never be satisfied with the solutions you currently use. Always be on the lookout for something better.
