There may never come a day when all application workloads run in a public cloud. However, the rate at which application workloads will shift towards public clouds appears to be poised to accelerate in the next few years.
LogicMonitor, a provider of IT monitoring tools delivered via the cloud, recently published a survey of 135 attendees of the recent AWS re:Invent 2018 conference. This survey finds that 63 percent of workloads are expected to be running in public clouds by 2020. That compares to 25 percent running on-premises, while another 12 percent will be running in a hybrid cloud. On average, survey respondents have 44 percent of their workloads running in a public cloud, while 46 percent run on-premises. In six years, however, only one-third of respondents says they expect to be running workloads both in the cloud and in an on-premises IT environment.
Interestingly, the survey finds reliance on hybrid clouds will not grow all that much in the next few years. On average, the survey shows only 12 percent of workloads will be running on hybrid clouds by 2020, up from 11 percent today. That lack of interest in hybrid clouds may be due to the nature of application workloads themselves or simply a lack of faith in the existence of tools to practically manage hybrid clouds being available in 2020.
The definition of what constitutes hybrid cloud computing is also evolving. Today, most workloads running in the public cloud are monolithic applications running on top of a virtual machine or deployed in a platform-as-a-service (PaaS) environment. The next generation of so-called cloud native applications are being built employing microservices architectures. The most common construct for building those applications is using containers deployed on top of a Kubernetes cluster. Kubernetes is an open source container orchestration software being developed under the auspices of the Cloud Native Computing Foundation (CNCF).
In most cases, today’s microservices are being deployed on top of traditional virtual machines or a PaaS. However, there is a nascent transition underway to deploy microservices either on micro virtual machines or bare-metal servers. To make matters even more interesting, many of these workloads will be dynamically invoking various serverless computing frameworks to handle child processes that only need to run for a few minutes at a time.
Emerging challenges for MSPs
The issue that managed service providers will need to quickly come to terms with is the growing diversity of application workloads running in the public cloud. Existing and emerging classes of platforms will be running side by side in the cloud all through the next decade, all of which creates a management challenge that most internal IT organizations are going to find daunting.
As each additional cloud-native computing framework gets employed, the odds those IT organizations will be looking to external providers of managed services starts to exponentially increase.
What will be interesting to see in the years ahead is how that shift will impact cloud service providers. The LogicMonitor survey of AWS re:Invent attendees unsurprisingly finds that 81 percent are using Amazon Web Services (AWS). Survey respondents employing Microsoft Azure and Google Cloud Platform (GCP) in production environments are 28 and 13 percent, respectively. But some are still experimenting with Microsoft Azure (19 percent) and GCP (26 percent). It would be fair for MSPs to assume that most IT organizations will be deploying workloads on multiple clouds for many years to come. The only real question is to what degree they might be able to unify the management of those hybrid clouds.
Regardless of how it all turns out, the one thing certain is that cloud computing is about to get a whole lot more complex. Those additional levels of complexity bode well for any MSP that can rise to that inherent management challenge.
Photo: Sushobhan Badhai / Unsplash