The multi-cloud paradigm is undergoing the kind of post-hype contrition that so many promising ideas experience once their initial glow starts to fade.  You might think, right, of course it is. Or, you might be wondering, “What is multi-cloud?” No matter. We will explain all. The purpose of this article is to review what multi-cloud was supposed to be and why it hasn’t matured in the way that has been articulated by many thought leaders. At the same time, we will point out how many of the best ideas contained in the multi-cloud concept can be realized through existing cloud technologies.

The idea for multi-cloud sprang from the reality that there are several major cloud platform providers, each with its own special features.  What if, the thinking went, organizations could avail themselves of the best from each of multiple cloud platforms?  That way, companies could avoid “vendor lock-in” by a single cloud vendor. They could reduce their vulnerability to a service outage by juggling workloads between cloud providers (easier said than done, but possible with the correct application architecture). They could diversify geographic reach, enhance agility, and so forth.  And, with a unified multi-cloud management toolset, it would be possible to stay on top of the whole multi-vendor structure through a single pane of glass.

Despite its promise and appeal, multi-cloud has not generally materialized in the IT world as people once predicted it would.  There are a number of reasons this is the case. Understanding multi-cloud’s failure to launch, though, provides an opportunity to understand how the cloud is evolving to help businesses achieve their strategic and operational goals.

  • It’s quite challenging to make it all work – Early multi-cloud platform offerings made it seem simple. You just plugged each cloud platform API into the multi-cloud controller and off it went.  But, it turned out to be a lot more challenging than that.
    • Monitoring of applications that could be moved from one platform to another was also an issue. Individual cloud providers offer their own dedicated management platforms, yet they are almost exclusively used to monitor individual services. Further, enterprises prefer to extend their existing monitoring tools versus learning entirely new products.
    • Integration of platforms into a single management layer was yet another concern, with difficulty growing as each provider introduces new features that need to be integrated.
    • And to top it all off, there is no such thing as an “industry standard API,” and many providers and technologies actually offer multiple APIs, and of course these change over time.
  • Security is a big deal – Securing assets in the cloud can take time, effort and expertise. Being able to shift those assets from provider to provider at will may sound like a good idea, but it’s a serious headache for IT. Controls and countermeasures tend to be platform-specific. Once set up and tested for audit purposes, they are sometimes best left alone.

All of this notwithstanding, it is possible to make multi-cloud work. But, it takes a lot of work.  Here’s the problem: Wasn’t avoiding work the whole point of the exercise? Ironically, enterprises that succeed are actually committed to fewer deep strategic relationships over the longer term. It doesn’t seem to be worth the effort.

The big insight that’s arisen over time is that it is possible to achieve the kind of portability pictured with multi-cloud, but at the application layer. A reasonable approach to consider is found in both Platform-as-a-Service (PaaS) and use of container technologies. By introducing a layer of abstraction – from the infrastructure or from the hypervisor respectively – companies can actually run applications wherever and whenever – thus reducing the challenge of vendor lock-in.

Hosted in the cloud, PaaS empowers companies to develop, run, and manage applications without the complexities associated with maintaining the supporting infrastructure. Containers also enable portability by eliminating the guest OS tax and by isolating the application within the container. Both provide the flexibility to easily move applications across infrastructure providers, yet in different ways. With a PaaS, companies will look for those supported by many providers (Cloud Foundry for example) as they will then be able to move their application another provider with the same underlying PaaS.

While the idea of multi-cloud held great promise, it’ may not prove as easy as once thought. Before taking that route, consider the functionality offered by PaaS and container technology to achieve the end goal. But don’t go it alone: Visit CenturyLink and see how we’re expanding on the possibilities of cloud. If you’re considering a multi-cloud approach to your IT infrastructure perhaps investigating the CenturyLink Cloud may be an option for you. Learn more.

This article was originally posted to Forbes Voice on May 12, 2016.

Leave a Reply

No comments yet