OpenShift al­tern­at­ives

OpenShift is the leading de­vel­op­ment platform for con­tain­er­ised ap­plic­a­tions based on Kuber­netes. A growing number of companies are using Kuber­netes as a cluster manager as a basis for operating and de­vel­op­ing con­tain­er­ised ap­plic­a­tions and services. In other words, Kuber­netes is a dis­trib­uted operating system for modern, container-based de­vel­op­ment.

The OpenShift platform was released by Red Hat – the maker of the pro­fes­sion­al Linux dis­tri­bu­tion ‘Red Hat En­ter­prise Linux’ (RHEL). The OpenShift al­tern­at­ive ‘Rancher’ has now been taken over by the tra­di­tion­al Linux provider SUSE. ‘Canonical Kuber­netes’, is another OpenShift al­tern­at­ive from an es­tab­lished Linux provider. Read on to find out more about these and other al­tern­at­ives to OpenShift.

OpenShift is a powerful system that combines a variety of func­tion­al­it­ies. In addition to Kuber­netes man­age­ment, it includes a dashboard to visualise clusters. Tools for mon­it­or­ing and logging, DevOps pipelines and a service mesh round out its functions that make it easy to set up dis­trib­uted ap­plic­a­tions and services. In summary, OpenShift and its al­tern­at­ives are Kuber­netes man­age­ment solutions.

Kuber­netes are the common feature of OpenShift and its al­tern­at­ives. But their functions extend far beyond managing con­tain­er­ised ap­plic­a­tions. In contrast to bare Kuber­netes, OpenShift and al­tern­at­ives control de­ploy­ment and operation of dis­trib­uted Kuber­netes clusters across the cloud and other in­fra­struc­tures. Thanks to user end-to-end au­then­tic­a­tion and de­ploy­ment of container images the systems are secure.

Our com­par­is­on does not include in­de­pend­ent al­tern­at­ives to Kuber­netes such as ‘Apache Mesos’ and ‘Azk’. These are described in our article on Kuber­netes al­tern­at­ives. Fur­ther­more, we shall not dive into Kuber­netes-as-a-Service (KaaS) offerings from major cloud providers. These can only be used as part of a cloud in­fra­struc­ture of the re­spect­ive provider, for example:

• IONOS Cloud Managed Kuber­netes
• Google Kuber­netes Engine (GKE)
• Amazon Elastic Kuber­netes Service (EKS)
• Azure Kuber­netes Service (AKS)
• IBM Cloud Kuber­netes Service (IKS)

Kuber­netes achieves a pre­vi­ously un­at­tain­able level of auto­ma­tion. However, Kuber­netes is a highly complex tech­no­logy. Not all or­gan­isa­tions have the ex­per­i­ence and capacity to run their own Kuber­netes in­fra­struc­ture. A ‘fully-managed’ Kuber­netes cluster, also known as ‘Kuber­netes-as-a-Service’ (KaaS) solves the problem: the provider manages a Kuber­netes cluster for the client in the pro­vider­'s cloud en­vir­on­ment. This allows the client to focus entirely on their core business.

A Kuber­netes man­age­ment solution is a man­age­ment layer based on Kuber­netes. Beyond container or­ches­tra­tion, the complete lifecycle of con­tain­er­ised ap­plic­a­tions can be managed. The Kuber­netes clusters used can be hosted on virtually any in­fra­struc­ture, from the public cloud to the company's own OpenStack en­vir­on­ment to ‘bare metal’ and edge devices. For the user, this creates func­tion­al­ity beyond Kuber­netes without tying the system to a specific in­fra­struc­ture provider.

For the purpose of com­par­is­on, we dif­fer­en­ti­ate between fully-managed and self-managed solutions. Some OpenShift al­tern­at­ives are available only as managed solutions. However, all enable operation of work clusters across in­fra­struc­ture bound­ar­ies.

The de­vel­op­ment of Kuber­netes man­age­ment solutions is in constant flux. Even though OpenShift al­tern­at­ives have similar functions, they often differ in the details. We summarise the most important features in the table below:

Below you’ll find a brief overview of OpenShift and select leading al­tern­at­ives. In each case, we address the ad­vant­ages and dis­ad­vant­ages of the al­tern­at­ive, their main dif­fer­ences compared with OpenShift, and common de­ploy­ment scenarios. The following Kuber­netes Man­age­ment Solutions are compared:

• Red Hat OpenShift
• SUSE Rancher
• Canonical Kuber­netes
• Mirantis Kuber­netes Engine (pre­vi­ously Docker En­ter­prise)
• VMware Tanzu (pre­vi­ously Pivotal)
• Platform9 Managed Kuber­netes
• Giant Swarm
• Portainer

OpenShift is the fron­trun­ner of Kuber­netes Man­age­ment Solutions created by Red Hat. The software is used to build dis­trib­uted, scaling ap­plic­a­tion and de­vel­op­ment en­vir­on­ments. Since the early days of Kuber­netes, Red Hat has con­trib­uted con­sid­er­ably to the de­vel­op­ment of the tech­no­logy.

The biggest advantage of OpenShift is the system’s range of functions. In addition to cluster man­age­ment, tech­no­lo­gies for the complete man­age­ment of the ap­plic­a­tion lifecycle are in­teg­rated, including de­vel­op­ment, rollout, operation, and main­ten­ance. Fur­ther­more, Red Hat provides its own container re­gis­tries, as well as a catalogue of operators.

The large feature set of OpenShift increases the com­plex­ity of the software. In­stalling and main­tain­ing OpenShift requires a team of experts, or support as part of a managed solution. To make matters worse, OpenShift’s Kuber­netes dis­tri­bu­tion differs sig­ni­fic­antly from the upstream version. Fur­ther­more, OpenShift only runs on Red Hat’s own operating system, Red Hat En­ter­prise Linux CoreOS (RHCOS).

OpenShift is also special in terms of the container images used. Due to the strict security settings, not all images from the freely available re­pos­it­or­ies can be used. Instead of the Helm charts known from Kuber­netes, OpenShift uses pro­pri­et­ary ap­plic­a­tion templates to build the in­fra­struc­ture. However, as of the launch of powerful OpenShift operators, an al­tern­at­ive mechanism now exists.

OpenShift is par­tic­u­larly suitable for im­ple­ment­ing hybrid cloud strategies and building and scaling the con­tain­er­ised ap­plic­a­tions required. Or­gan­isa­tions that are already part of the Red Hat ecosystem will benefit from using the software.

Rancher is a long-standing OpenShift al­tern­at­ive that was acquired by SUSE at the end of 2020. Rancher provides a powerful yet flexible Kuber­netes man­age­ment platform. The software fa­cil­it­ates man­age­ment of multiple Kuber­netes clusters across different cloud en­vir­on­ments. Range of functions includes pro­vi­sion­ing, mon­it­or­ing, and logging.

Besides the ob­lig­at­ory cluster man­age­ment, Rancher in­teg­rates a number of in­ter­est­ing tech­no­lo­gies. For example, the ‘Rancher Kuber­netes Engine’ (RKE) is a Kuber­netes dis­tri­bu­tion that is published as open source. It offers a minimal K3s dis­tri­bu­tion for use on edge and IoT devices. This can also be used to implement single node clusters. Rancher features access controls according to the ‘Open Policy Agency’ (OPA) standard and supports the ‘Longhorn’ storage engine.

Rancher is par­tic­u­larly well suited for managing multiple Kuber­netes clusters with a cent­ral­ised interface. User rights can be assigned centrally across multiple clusters.

Canonical Kuber­netes is another Kuber­netes man­age­ment solution from a well-known operating system provider. Ubuntu Linux, developed by Canonical, serves as the found­a­tion for Kuber­netes dis­tri­bu­tions from major cloud providers such as Google Kuber­netes Engine (GKE), Amazon Elastic Kuber­netes Service (EKS), and Azure Kuber­netes Service (AKS). Thus, with Canonical Kuber­netes, users can build clusters that span cloud bound­ar­ies.

The biggest ad­vant­ages of Canonical Kuber­netes are flex­ib­il­ity and low cost of de­ploy­ment. ‘Charmed Kuber­netes’ is an exciting operator-based approach to build in­fra­struc­tures. ‘Charms’ is a col­lec­tion to seam­lessly integrate and manage Kuber­netes, container and VM-based ap­plic­a­tions across hybrid cloud en­vir­on­ments. Canonical Kuber­netes uses Ubuntu as operating system.

Canonical Kuber­netes is suitable for a wide range of hardware. Processor ar­chi­tec­tures x86, ARM, IBM POWER, and IBM Z are supported; GPUs can be used in addition to CPU. For use on desktop and edge devices, there is the minimal ‘MicroK8s’ Kuber­netes dis­tri­bu­tion, which is par­tic­u­larly resource-efficient.

Mirantis purchased ‘Docker En­ter­prise Edition’ from Docker Inc. in 2019, and since then the tech­no­logy has been re­launched as ‘Mirantis Kuber­netes Engine’. The software is rooted in the Docker universe. In addition to Kuber­netes, Docker Swarm is a supported cluster manager.

The Mirantis Kuber­netes Engine occupies an in­ter­est­ing niche. For example, in addition to Linux, it also supports Windows to run on nodes. In­ter­est­ingly, Linux and Windows nodes can be mixed seam­lessly, and the same applies to nodes with Kuber­netes or Docker Swarm. The product is aimed at larger busi­nesses. Thus, it un­for­tu­nately lacks a vibrant community of de­velopers and the free trial version requires re­gis­tra­tion.

The tech­no­logy is par­tic­u­larly suitable for larger or­gan­isa­tions with a strong focus on the developer ex­per­i­ence. If an in­fra­struc­ture based on Docker Swarm or Windows nodes already exists, it is re­l­at­ively easy to extend it with Kuber­netes using the Mirantis Kuber­netes Engine.

VMware Tanzu is a Kuber­netes platform from vir­tu­al­isa­tion spe­cial­ist VMware. The product resulted from the ac­quis­i­tion of Pivotal by VMware at the end of 2019. With VMware Tanzu, Kuber­netes clusters can be built on VMware’s vSphere in­fra­struc­ture. The in­teg­rated ‘Tanzu Mission Control’ is a cent­ral­ised interface for managing dis­trib­uted Kuber­netes clusters. In addition to func­tion­al­it­ies for de­vel­op­ment and operation of con­tain­er­ised ap­plic­a­tions, mon­it­or­ing and logging, network and storage man­age­ment are included.

The ad­vant­ages of VMware Tanzu are primarily its in­teg­rated tools. The ‘VMware Tanzu Ap­plic­a­tion Catalog’ contains a well-main­tained catalogue of open source com­pon­ents and ap­plic­a­tions while ‘Harbor’ is a dedicated container registry to secured different build artifacts by means of role-based access controls. The container images stored in Harbor are checked for security holes and vul­ner­ab­il­it­ies and marked as secure. The focus when using VMware Tanzu is clearly on the de­ploy­ment of vSphere in­fra­struc­ture. Support for other cloud platforms is less soph­ist­ic­ated.

VMware Tanzu is par­tic­u­larly suitable for companies that already know their way around the vSphere in­fra­struc­ture. In addition to the com­mer­cial version, a free version is available for Linux, Windows and Mac – the ‘VMware Tanzu Community Edition’.

Platform9 was founded by former VMware engineers. The core concept of the platform is to run es­tab­lished open-source software like Kuber­netes under a software-as-a-service (SaaS) model for customers. This makes it possible to centrally manage Kuber­netes clusters dis­trib­uted across different cloud en­vir­on­ments. Platform9 takes over the main­ten­ance of the Kuber­netes in­stall­a­tions, which reduces op­er­a­tion­al com­plex­ity on the customer side.

The biggest advantage of Platform9 is its ‘fully managed’ operating model. Clients no longer have to deal with operating system or Kuber­netes updates. Other positive features are its clear pricing structure, extensive doc­u­ment­a­tion, and a free trial version. Even though the web interface is a bit outdated, it includes plenty of standard features: the ‘Pro­meth­eus’ software con­tinu­ously checks the health of clusters, and logs are collected and analysed centrally. Ap­plic­a­tions available in public Helm re­pos­it­or­ies can be installed without any problems.

As a special niche, Platform9 focuses on tele­com­mu­nic­a­tion service providers that unite large quant­it­ies of edge devices in Kuber­netes clusters. Fur­ther­more, the platform makes it possible to run ‘legacy’ apps based on virtual machines in parallel to modern, con­tain­er­ised ap­plic­a­tions.

Giant Swarm is a ‘fully-managed’ Kuber­netes platform. The software takes over the man­age­ment of the Kuber­netes en­vir­on­ment, ir­re­spect­ive of the type of un­der­ly­ing cloud in­fra­struc­ture used. Clients can thus focus on ap­plic­a­tions and services run on Kuber­netes. The man­age­ment and worker clusters can be operated on Amazon AWS, Microsoft Azure or ‘on-premises’ as part of a client’s existing computing in­fra­struc­ture.

The platform manages and optimises the entire lifecycle of con­tain­er­ised ap­plic­a­tions and is quickly de­ploy­able. Unlike OpenShift, a Giant Swarm in­stall­a­tion can be set up within a day. Since an un­mod­i­fied Kuber­netes dis­tri­bu­tion is used, it is re­l­at­ively easy to move parts of the in­fra­struc­ture between cloud en­vir­on­ments. This reduces the risk of vendor lock-in. On the negative side, it is no­tice­able that there is no uniform pricing model. Instead, the price is tailored according to the client’s needs.

Giant Swarm is par­tic­u­larly suitable for con­trolling dis­trib­uted mi­croservice ar­chi­tec­tures. This allows resilient, agile systems to be built and scaled. The software takes over the complex man­age­ment of the un­der­ly­ing in­fra­struc­ture.

Unlike the afore­men­tioned OpenShift al­tern­at­ives, Portainer follows a different concept. It is not a fully-managed platform or Kuber­netes dis­tri­bu­tion of its own. Instead, it uses a graphical user interface that unifies the man­age­ment of existing Kuber­netes de­ploy­ments. DevOps teams have the ability to centrally manage, configure, and secure multi-cluster en­vir­on­ments. At the developer level, de­ploy­ment, man­age­ment, and troubleshoot­ing of con­tain­er­ised ap­plic­a­tions is sim­pli­fied.

The main advantage of Portainer is the flex­ib­il­ity of the software. In addition to Kuber­netes, Docker Swarm and Docker can be used to manage clusters and con­tain­ers. Portainer is based on open-source software and is offered in a freely available community version as well as a paid version with en­ter­prise support. The software can be installed in cloud en­vir­on­ments, on edge devices, or within a client’s existing computing in­fra­struc­ture. Since Portainer is re­l­at­ively new, the doc­u­ment­a­tion leaves much to be desired.

Portainer is par­tic­u­larly suitable for cent­ral­ising and unifying the man­age­ment of existing cluster in­fra­struc­tures. If Docker or Docker Swarm-based systems are used in addition to Kuber­netes, they can be managed with ease.