IT leaders are deploying an growing variety of containers to modernise purposes, run business off-the-shelf software program, and deal with synthetic intelligence (AI) and superior analytics workloads.
Whereas containerisation is related to a cloud-native microservices structure, it is usually used to host conventional enterprise purposes, resembling when organisations wish to cut back their VMware property and choose Kubernetes to assist each cloud-native and conventional purposes. As a consequence, extra organisations are having to work out how greatest to deal with persistent storage with Kubernetes (see Acid versus Base field).
A Dimensional Analysis ballot of 519 IT decision-makers at organisations with 500 or extra staff for the Portworx Voice of Kubernetes report 2026 discovered that 71% of organisations plan to modernise and/or migrate their digital machines to Kubernetes.
Pods and stateful purposes
Containerisation began out as an strategy to creating fashionable, cloud-native purposes that have been “stateless”. Nevertheless, the Portworx report said that 58% of IT decision-makers plan to run stateful purposes resembling databases on Kubernetes.
Whereas Linux containers – often known as pods within the Kubernetes world – can instantly entry storage, the disk information in a container have a brief lifespan and are deleted when the container goes down.
how Docker containers was deployed, analyst agency Gartner’s Learn how to deploy and function containers and Kubernetes report notes that these containers didn’t learn info, other than the parameters offered of their configuration information, and they didn’t retailer their state outdoors the operational scope or lifetime of a single container occasion.
This has ramifications for IT groups tasked with reengineering enterprise purposes to make them cloud native, the place they’re containerised and managed by Kubernetes.
Gartner is now seeing curiosity in utilizing containers to deploy stateful purposes, which suggests information will be accessed past the lifetime of a container. Stateful containers are additionally protected with backup and restore processes, and as Gartner notes, information could also be shared by a number of containers.
Containerisation of edge compute
One motive that extra consideration is being positioned on persistent storage now could be the info storage necessities of edge computing, which can want to deal with intermittent community connectivity or function standalone.
In a weblog revealed in 2022, Kevin Pitstick and Jacob Ratzlaff, software program engineers on the Carnegie Mellon Institute, wrote about the advantages of utilizing containerisation on the edge.
Pitstick and Ratzlaff wrote that containerisation permits restoration and continued operation underneath fault circumstances. If a containerised edge utility is engineered utilizing a microservice structure, one other good thing about containerisation is that if a container utility crashes, solely a single functionality goes down fairly than the entire system.
They famous that Kubernetes can benefit from digital IP addresses to permit failover to a backup container if the first container goes down, and orchestration can be utilized to redeploy containers mechanically for long-term stability.
In accordance with Pitnick and Ratzlaff, containers will be simply unfold throughout a number of edge methods to extend the possibility that operations will proceed if one of many methods will get disconnected or destroyed.
China’s first orbit interconnected house computing constellation, led by Zhijiang Laboratory, makes use of CubeFS, a Cloud Native Computing Discussion board (CNCF) venture labeled as “graduated”, because the storage base of its space-based distributed working system to handle information distributed throughout a number of satellites. CubeFS was initially developed by Chinese language e-commerce large JD.com. Together with the resilience wanted in software program that operates in house, CubeFS has been optimised to deal with giant volumes of small information effectively.
Such performance is vital for scaling methods, one thing that’s gaining extra consideration as IT decision-makers take into account the IT infrastructure necessities for synthetic intelligence (AI).
Approaches to persistent storage
In a 2024 survey of 372 organisations, the CNCF discovered that organisations with just some expertise of cloud-native computing and containerisation have been extra prone to discover storage a problem (15%) than those who take into account themselves mature cloud-native organisations (13%). Apparently, the organisations with the least concern about storage have been those who had not but began their cloud-native journey (11%).
Whereas it’s doable to deploy stateful purposes that require persistent storage utilizing public cloud companies like Amazon Easy Storage Service (Amazon S3), such companies are typically much less versatile than open supply Kubernetes persistent storage.
Previous to Portworx, we weren’t in a position to assist our database workloads on Kubernetes. Now that now we have Portworx on the core of our infrastructure, we offer our shoppers with the last word flexibility and velocity for his or her most demanding persistent storage wants Jeroen van Gemert, KPN
On the KubeCon convention in Amsterdam in March, SuSE mentioned an replace to its Longhorn persistent storage venture to assist high-performance capabilities wanted for AI. Roy Illsley, principal analyst at Omdia – who attended the convention and was additionally on the Nvidia GTC occasion in San Jose – says cloud-native persistent storage is especially related for AI workloads.
“It’s most likely turning into extra related as individuals are realising that if we wish to do AI and machine studying, and we wish to use containers or cloud-native know-how, then we’re going to want persistent storage,” says Illsley.
SuSE sees persistent storage as key to the profitable modernisation of purposes that may benefit from containers and orchestration. It positions Longhorn as a product to assist IT infrastructure groups handle efficiency issues, resembling databases slowing down and purposes experiencing a rise in latency. In accordance with SuSE, IT tends to handle such efficiency bottlenecks by over-provisioning high-speed storage. Longhorn is classed as an incubating venture by CNCF, which suggests it’s secure sufficient for manufacturing environments.
SuSE says its Longhorn V2 Knowledge Engine avoids utilizing the Linux kernel storage stack, and has been engineered to keep away from the context switching overhead inherent when conventional storage pipelines are used. SuSE says the brand new structure permits storage operations to run on devoted central processing unit (CPU) cores to unlock increased throughput and decrease latency.
It claims that early benchmarking has proven that Longhorn V2 achieves a two to 4 occasions enchancment in write efficiency and as much as two to a few occasions quicker random learn efficiency in contrast with the earlier model. Together with different enhancements, SuSE stated the brand new model gives 10 occasions the efficiency of Longhorn V1 in sure situations.
Portworx, which was acquired by Pure Storage in 2020 and is now often known as Portworx by Everpure, is a business different to SuSE Longhorn, constructed on open supply applied sciences. At KubeCon, it unveiled new capabilities for Portworx Enterprise to assist organisations operating enterprise-level digital machines (VMs) and containers in cloud, hybrid and on-premise environments. The corporate says these updates permit organisations to embrace fashionable virtualisation, thereby accelerating provisioning, simplifying automation and saving time.
Dutch telco KPN has been utilizing Portworx for numerous years to assist software program that requires persistent storage when it’s containerised, as Jeroen van Gemert, DevOps engineer at KPN, explains: “Previous to Portworx, we weren’t in a position to assist our database workloads on Kubernetes. Now that now we have Portworx on the core of our infrastructure, we offer our shoppers with the last word flexibility and velocity for his or her most demanding persistent storage wants, from databases like Redis and ElasticSearch to different excessive IO [input/output] workloads.”
Persistent storage for digital sovereignty
Together with containerisation of edge computing purposes, AI and the migration of workloads away from VMware, the rise in demand for information sovereignty means IT leaders are asking for stringent management over the place their enterprise information is saved.
SAP, for example, is constructing a digitally sovereign 120PB (petabytes) storage spine for Europe, primarily based on the open supply Ceph software-defined distributed storage platform and Rook, a storage orchestrator for Kubernetes, which offers a administration layer. Rook is one other CNCF venture. Its CNCF classification is graduated, making it appropriate for manufacturing environments.
Because the containerisation of stateful purposes turns into extra frequent, Gartner recommends IT decision-makers take into account a wide range of approaches for offering information persistence to containerised purposes. Whereas offering persistent information storage for purposes that run on a singular host is easy, the analyst agency notes that with purposes clustered throughout a number of hosts for the aim of excessive availability or scalability, orchestrating the state of this information will be tougher.
