Why multicloud platforms fail and how to prevent that
Originally published at TechBeacon
Originally published at TechBeacon
The right multicloud management strategy starts with a thorough understanding of why you need to operate in more than one cloud.
Your cloud management platform needs will vary depending on whether you are already using multiple clouds, whether you have legal or historical business requirements, and whether you are pursuing multicloud for strategic business reasons.
These reasons include using best-of-breed tooling in each cloud or having servers located as close to your users as possible to reduce latency.
You can read more about all the ways organizations end up adopting multicloud in an older post I wrote.
At the strategy stage, one common mistake is not fully considering which scenario best fits your organization's situation. This problem has implications for things such as:
- Whether a top-down or bottom-up approach to platform development is most appropriate.
- What functionality you will need in the multicloud management platform.
- Which stakeholders need to sign off on the platform and who would decide if it's a success or failure.
Generally speaking, it is more complicated to build a management platform for existing setups than it is to start from scratch. But as long as you're aware of how complicated things can get, it is easier to plan for that complexity at the architectural phase.
Regardless of whether you are building from scratch or trying to wrangle a mess of existing applications running in multiple cloud environments, building the right platform will help you tame the complexity and create a more consistent experience for everyone involved, from developers to business stakeholders.
Most organizations, however, fail on their first attempts at a multicloud platform. Here's why and how to avoid these common mistakes.
The four most common strategic mistakes
As companies start to pursue multicloud, they make four strategic mistakes that occur regardless of why they decided they needed the platform in the first place.
1. "Let's just do it ourselves. How hard can it be?"
Building a multicloud management platform sounds like an interesting challenge to a lot of engineers. The problem is that the types of engineers with the skills to successfully set up multicloud usually have other responsibilities and can't spend months focused solely on building a new platform.
It is also more complicated than most engineers realize. And, eventually, one or more of the people who built the platform will get a better job offer and leave. The organization will be left with a platform that no one knows how to maintain.
2. "Let's just choose the right silver bullet."
The second mistake people make is thinking that there is one correct technology to manage multiple clouds. The truth is that there are many options.
The best approach is to take advantage of several of them, because none will cover every use case you want on its own. You have to be willing to adapt and modify everything. Every successful multicloud management platform is a custom one.
3. "Just let the ops people choose."
At the end of the day, whatever multicloud platform you choose or build is going to have to work for all of your stakeholders. This includes technical developers, perhaps less technical engineering managers, and even less technical business leaders.
Dev teams, ops teams, and infosec teams are going to have different priorities and different deal breakers. If any of the stakeholders don't like the platform you build, they won't use it and suddenly developers will be spinning up AWS services directly from the AWS portal. Nothing will actually make it into the multicloud platform, rendering it useless.
4. "My friend at company X bought vendor A's solution and is using it off the shelf. Let's just do that."
While completely rolling your own solution nearly always ends in failure, that doesn't mean you should treat multicloud platforms like interchangeable widgets. Each organization's needs are unique, and what works in one place will not necessarily work elsewhere.
Even in the best-case scenario, you should expect to customize the platform to some extent. It is not something that you can realistically expect to just plug-and-play and get everything your organization needs.
Avoid failure with the right approach
Multicloud management platforms are complex, and you will have failures. You want to end up with a successful solution after many minor failures instead of with a spectacularly expensive failure of what was supposed to be a finished product.
The best way to do this is to take small steps and iterate constantly, with continual feedback from all of your stakeholders.
You will need to customize the right mix of tools, workflows, and functionality and make sure everyone who needs to use the finished platform buys into the process and has a voice in its direction.
Ultimately, the right mix of vendors, open source projects, and homegrown customizations will depend on the unique needs and skill sets of every organization.
Approach the search for a multicloud management platform with clarity about why you need to be in multiple clouds and with organizational self-awareness about your strengths and weaknesses. You'll be more likely to end up with a solution that helps you reach your strategic goals while taming the inherent complexity of operating across environments.