Kubernetes and storage in the enterprise: What the analysts say

How widespread is Kubernetes in the enterprise?

There is no doubt that Kubernetes adoption is growing, especially in larger organisations, but analyst numbers vary, and there are suggestions that data on the appetite for Kubernetes might overtake its real-world take-up.

Gartner, for example, believes less than half of global organisations ran containerised applications in production in 2023, but that 95% of global organisations will do so by 2029. Moreover, 35% of enterprise applications will run in containers – more than double today’s figures – and 80% of commercial software will be offered in a containerised version.

At ESG, analyst Simon Robinson cites a 2023 research study that found 67% of firms used containers for production apps. “Almost all of the rest are planning to do so in the future,” he says.

On the storage side, again in 2023, 45% of organisations used persistent storage with production apps, and 37% ran test and development with storage. But not all analysts are as positive.

Research by CCS Insight, sponsored by Red Hat, found that back in 2021, 90% of respondents were interested in the technology. “Just over 70% [were] getting ready to deploy into production or working on proof-of-concept projects,” explains Bola Rotibi, CCS’s chief of enterprise research. “My prediction is that its adoption will continue, but not at such a fast rate as implied by some in the market.

“Some surveys put Kubernetes’ share of the container technology market at around 24%, with the US being seen as the biggest user,” she says. “However, despite the positivity around Kubernetes adoption rates at KubeCon in March this year, there was a lot of discussion around the realities of Kubernetes adoption that it is not as widespread as some industry reports may suggest.” 

Rotibi highlighted one organisation that has almost 20 times as many VMs as it has Kubernetes-based workloads as a benchmark for how things are in some cases.

What are the obstacles to Kubernetes deployment?

The obstacles to Kubernetes take-up will hardly come as a surprise. Containers are complex, and can be difficult to move to production.

Gartner analysts Arun Chandrasekaran and Wataru Katsurashima cite “a lack of adequate skills and mature DevOps practices to operationalise and succeed with large-scale production deployments”. This is despite a growing number of use cases.

But it is the complexity of Kubernetes and the range of different ways it can be deployed that still hold it back. It’s also worth noting that stateful Kubernetes deployments are still somewhat less mature than the original, stateless concept.

“In our research, we’ve seen complexity challenges being the number one reason organisations struggle to adopt cloud-native practices, including Kubernetes,” says Jon Collins, vice-president of engagement at analyst GigaOm. “There’s a mushy middle, as applications migrate from single-stack legacy, through virtualised and lift-and-shift cloud, to architectures and approaches that are rooted in cloud principles.”

He adds that complexity management is a challenge when it comes to scaling up any cloud-native applications, with a lack of clarity around containerised applications that increases security and other risks.

“Overall, the deployment/management process needs to simplify, especially when running at scale,” says ESG’s Robinson. “Kubernetes is still pretty complex. Also, in larger organisations, there are often many Kubernetes teams deploying apps, often using different tools and methodologies, often in a range of locations. This is inefficient and introduces risk for the organisation.”

For IT leaders, the challenge is to square the circle of running complex, enterprise workflows in a system designed to be stateless and have a very light resource footprint.

Kubernetes, storage and data protection

Data storage remains an area where organisations struggle with Kubernetes. Persistent storage, backup and recovery still feel like they have been grafted onto the Kubernetes project.

“One common misconception about Kubernetes is the idea that applications that run on the platform must be designed to be stateless,” say Gartner analysts Fintan Quinn and Tony Iams. “In reality, most applications and projects that run on a Kubernetes cluster will eventually need to store or access data.”

Stateful applications and workloads use persistent volumes (PVs) to store data, with the PV able to retain data even if the pod using it stops working. But, as Quinn and Iams suggest, this needs careful design to avoid data loss.

“For storage specifically, there are a range of challenges, but the key ones are managing quality of service, managing across hybrid environments, compliance, cost and performance,” says Robinson. “When you fire up several thousand containers at once, you need a lot of fast storage!”

When it comes to backup and recovery, chief information officers need to ensure Kubernetes environments are actually part of the backup or disaster recovery plan. Often, they are not.

“Running stateful workloads within your Kubernetes cluster is okay, but it’s crucial to adapt backup and recovery processes to include application data,” say Quinn and Iams.

At CCS Insight, Rotibi points out that better “monitoring, observability and visibility” is needed in Kubernetes environments. Simplifying deployment and management will also make Kubernetes more robust.

And over time, we are likely to see better support for Kubernetes in conventional backup and recovery tools, as well as more robust backup options in Kubernetes management environments.

“It’s still early days, but the main trend is growing adoption of stateful apps,” says Robinson. “Data needs storing and protecting much like for traditional apps. This is part of the ‘shift left’ trend to get more traditional enterprise data services baked into the process earlier.”