As profound an impact as cloud computing may have had on the delivery of IT services, managed service providers would do well to note that the shifting of virtual machines from an IT environment running on premise to shared public infrastructure represents a modest innovation from a technical perspective.
In fact, some ten years since public clouds were first adopted, it’s only now with advances such as Docker containers and serverless computing that we are about to witness IT being fundamentally transformed.
Impact of containers and serverless computing
Docker containers are a lightweight approach to packaging application code and the underlying primitives needed to run that code in a single payload that can be deployed on top of a virtual machine or on a bare metal server. Most containers today run on top of virtual machines because IT organizations typically don’t have the IT management tools and expertise needed to run them on bare metal servers. But eventually most containers will wind up running on bare metal servers because the economics in terms of server utilization rates will be too great to ignore. More than five times as many containers can run on a bare metal server than can be deployed on a virtual machine.
Today, containers have been used mainly to support stateless applications. It’s only relatively recently that containers have started being used to support stateful applications that require access to persistent storage. And, the overwhelming majority of enterprise applications tend to be stateful rather than stateless.
In the meantime, serverless computing frameworks such as AWS Lambda are starting to gain traction. Despite the naming convention, serverless computing still makes use of servers. The difference is that they're built on top of event-driven frameworks that make it possible to invoke additional infrastructure resources on demand without needing to interrupt the application to make it aware of those additional resources. Given that approach, it turns out that serverless computing frameworks are ideal for stateless applications. In effect, there’s now a fork in the road ahead that is likely to see stateless applications moving in the direction of serverless frameworks, while stateful applications start to move over to containers.
As all this occurs, reliance on virtual machines in the cloud will start to taper off. In fact, the cost of running application workloads in the cloud on either bare metal servers or using serverless frameworks will drop considerably as server utilization rates increase significantly.
Opportunities and challenges for MSPs
Going forward, MSPs will soon find themselves supporting three different approaches to cloud services. One or two of these approaches may supplant the other. But for all practical purposes, some form of all three will be employed well through the end of the decade. The challenge MSPs will face is that it will be difficult to centrally manage all three approaches from a single management console, especially across public and private cloud computing environments. MSPs will also soon find themselves being required to invest in additional expertise to serve markedly different forms of cloud computing.
The good news is that internal IT organizations are likely to face the same challenges without nearly the same level of resources. The more complicated things become, the better off most MSPs will be. In fact, at this rate of innovation the number of workloads running in the cloud is likely to increase by several orders of magnitude. As the ratio between the number of application workloads running in the cloud to internal IT staff available to manage them increases, more organizations than ever will be forced to look outside for external expertise to help them cope with what is increasingly becoming a much more diverse world of cloud computing.