<img alt="" src="https://secure.bomb5mild.com/193737.png" style="display:none;">

Turbonomic Blog

The Truths and Myths of Multi-Cloud

Posted by Eric Wright on Jun 17, 2021 12:11:54 PM
Eric Wright
Find me on:

It's no secret that the cloud has been a disruptive technology, and as it is still an emerging field of computing in relative terms of the age of some services, there are many misconceptions about what multi-cloud infrastructure really means. Whether you're considering going all-in on one provider or spreading your data across providers for higher availability, we'll cover some of the truths and myths behind this exciting new way to build out your IT infrastructure.


I had a great discussion recently with Chris Psaltis, founder of Mist.io, and we talked about a number of these topics in-depth. Here are the hot button items that are ones that come up the most often.

Myth #1 - Multi-Cloud is a Strategic Advantage

This one may be considered by a choice view as a truth, but there are a limited and specific set of use-cases that allow for multi-cloud to be a strategic advantage. Let's start with a definition:

Strategic: Carefully designed or planned to serve a particular purpose or advantage.

A dominant percentage of multi-cloud environments are a result of necessity rather than strategy. We find many companies will acquire a product or platform that is limited to one cloud (public or private) and goes counter to their main adopted cloud platform. A popular example is acquiring a company which built their infrastructure on a AWS when your organization was heavily using Azure already.

Real use-cases for making multi-cloud strategic are often cited including single applications that span multiple clouds for resiliency, using multiple clouds as a competitive bargaining advantage for long-term contract negotiations with each cloud provider, and removing "lock-in".

The reason this idea is more myth than truth is that there are very few organizations really using multiple public clouds as a purely strategic advantage rather than just out of necessity due to specific technological requirements.

Myth #2 - VMware Cloud is Multi-Cloud

This may be an unpopular opinion. VMware Cloud platforms have been strategic and incrementally innovative for many companies who have been relying on the core hypervisor for years. Being able to move your workloads to a different underlying infrastructure in any data center was a huge advantage.

VMware had many fits and starts with public cloud offerings and landed with the most success so far with VMware Cloud on AWS. This does present a way to remove the need to spin up the data center itself, but it's really cloud-adjacent more than truly multi-cloud.

There are many advantages to leveraging VMC on AWS (and Azure and GCP now) that include leveraging the adjacent services that are cloud-specific. It's very valuable to be able to run your existing front-end app servers on VMware hypervisors and leverage back-end data services like AWS RDS, AWS RedShift, etc.

It has been amazing to see companies use VMC to reduce the need to launch their own physical data center and also using it for business continuity and disaster recovery, but the truth is that it is just the same platform using a cloud provider as a data center provider more than VMC itself being a true multi-cloud platform.

Myth #3 - Multi-Cloud is Easy

It's not always easy to move from one cloud to another. Some apps and workloads are very sensitive to where they run, so moving them may require a lot of re-architecting work. And then there is the problem of data storage orchestration - if you want your data available in two clouds at once for example, your teams have both a tooling and a data access latency challenge in front of them.

Even simple applications with simple data patterns have a surprising number of external implications. These include things like identity and access management, security, logging, provisioning and operational scripts and tooling, and much more. That simple app that can be run on two clouds means your teams will need a broader skill set, cloud-agnostic tooling, and things get complicated quickly when applications spanning multiple clouds begin to fail. Where do you start looking when things go wrong?

Truth #1 - Multi-Cloud is Widely Adopted

Despite the myths we've listed above, there are a massive number of organizations using multi-cloud as an operational pattern in one way or another already. That's because for many organizations, the flexibility of having a multi-cloud option is just too compelling to ignore.

Despite all this complexity, managing an application environment across multiple clouds can be surprisingly simple if you're using the right tools. There are many cloud-agnostic platforms that will provide the right abstraction from the underlying infrastructure. This translation from infrastructure-specific to intent-driven is where we will see a significant amount of innovation in the coming years.

When your company acquires a product or even another company with cloud-specific requirements outside of your current cloud portfolio, it is probably not strategic to throw all of your engineering at re-platforming and rebuilding your application. This lets teams just focus on a long-term strategy and to maintain and manage the multi-cloud environment as an operational reality.

Truth #2 - Kubernetes has Unlocked Multi-Cloud Like Never Before

Multi-Cloud is not a new concept, but the ability to have Kubernetes orchestrate all of these different cloud providers in a loosely coupled manner with API abstraction has unlocked multi-cloud like never before. The success of OpenShift, as an example, gives the ability to run a common underlay and abstract a lot of the operational complexity away.

There are still some things like identity management, billing, logging, security, and others that require additional care and feeding. That said, you have one console to go to for managing most of the operations and you can spin up environments with a click or an CLI command rather than navigate the complexity of cloud-specific consoles.

Kubernetes also solved the problem which really matters for applicaiton builders: a common developer experience. Using a common approach to build and deploy the application means we can get that elusive "velocity" that companies have been seeking for decades. Now you truly can write once, run anywhere.

Truth #3 - Regional Availability Matters

Regional availability matters and we need to make sure that we are providing services or products on each region where our customers reside, so they don't have to worry about latency or lack of service availability when trying to execute their tasks. The more regions there are available for access means the better off companies will be because it can eliminate any single point of failure risk as well as provide redundancy if one regional site goes down.

Multi-cloud means you can put services closer to the customer. Hybrid deployments are hugely popular for this reason. You can have the bulk of your application infrastructure in your own data centers and then host a subset of your apps in a public cloud provider. The nominal cost of the complexity is outweighed by the customer satisfaction and application performance advantage.

Truth #4 - SaaS and PaaS Simplify Multi-Cloud Challenges

While we are all wondering whether multi-cloud is strategic, the real strategic advantage is being unlocked by leveraging SaaS and PaaS platforms. Now you can use DBaaS like AWS RDS and let your DBAs focus on data structure instead of loading up SQL servers on IaaS instances.

You can use platforms like Salesforce instead of building your own bespoke CRM. You can test out SaaS and PaaS without the upfront investment in development and IaaS setup that was needed before. There are so many possibilities with on-demand access to many products and platforms.

In Conclusion

Multi-cloud is here, and while it has strategic advantages among the patterns we've talked about above, it is rarely strategic to specifically build a multi-cloud approach. A strategic advantage is to leverage the best of breed for specific services, to lean into your teams for their own comfort and skills to choose the right platform, and to accept that being single cloud is also not necessarily a strategic choice.

It’s weighing the full cost of operations (people, process, technology) as you decide on how to build and continue to operate your infrastructure. After all, it’s really all about the applications and what you do with them. That’s where the strategic value comes in.

Our team here at Turbonomic have been doing a lot of open field research on this topic and have a lot of great data and results to share if you want to see for yourself where these myths and truths come into play. You can head over to https://turbonomic.com/multicloud and get the full survey results from our annual State of Multicloud survey that unpacks some very interesting trends.

The multi-cloud future is here and it's wonderful in the potential it creates for us to do strategic things. But multi-cloud is not the strategy itself.

Topics: multicloud, multi-cloud

Subscribe Here!

Recent Posts

Posts by Tag

See all