IBM Maximo on On-Premises
IBM Maximo Application Suite 9 on Red Hat OpenShift in your own data center, on VMware vSphere or other existing virtualization, including disconnected installs for plant networks and data residency
Overview
IBM Maximo on-premises in MAS 9 means Red Hat OpenShift running on infrastructure you already own, most often VMware vSphere. OpenShift supports four vSphere install methods: installer-provisioned, user-provisioned, the Agent-based Installer, and the Assisted Installer, and Red Hat documents disconnected installs for the installer-provisioned, user-provisioned, and Agent-based paths. A MAS license includes an OpenShift subscription, and the MAS CLI installs the suite from a static IBM operator catalog. MAS needs both a ReadWriteOnce and a ReadWriteMany storage class, which on vSphere usually means OpenShift Data Foundation or the vSphere CSI driver with vSAN file service. The installer can provision Db2 and MongoDB inside the cluster or connect to databases you already run. For sites that cannot reach the internet, the MAS CLI mirrors IBM and Red Hat images to a private registry and configures the cluster to pull from it. We design, build, and upgrade these environments, including moves from Maximo 7.6.1.
What teams weigh before running Maximo on On-Premises
What on-premises means in MAS 9
Maximo 7.6 ran on WebSphere and a database server. MAS 9 is a set of OpenShift operators, so on-premises now means running an OpenShift cluster. IBM ended regular base support for Maximo Asset Management 7.6.1.x on 30 September 2025, and MAS 8.10 and 8.11 standard support ended on 30 April 2026. Extended and Sustained Support can be purchased for 7.6.1.x, but the long-term path is MAS 9.
Reference architecture
Red Hat OpenShift Container Platform on vSphere virtual machines, installed with installer-provisioned infrastructure, user-provisioned infrastructure, the Agent-based Installer, or the Assisted Installer. Red Hat also documents three-node compact clusters on vSphere for smaller footprints.
Control plane, worker, and optional infrastructure node VMs sized for the MAS applications in scope. Dedicated storage nodes are common when OpenShift Data Foundation runs in the cluster, and GPU worker VMs can be added for Maximo Visual Inspection.
MAS requires both a ReadWriteOnce and a ReadWriteMany storage class. OpenShift Data Foundation provides ocs-storagecluster-ceph-rbd and ocs-storagecluster-cephfs. The vSphere CSI driver provides the thin-csi class, which supports ReadWriteMany only when vSAN file service is configured. NFS is another documented option.
The MAS CLI can provision Db2 and MongoDB inside the cluster or connect to databases you bring. Maximo Manage also supports Oracle Database and Microsoft SQL Server, which suits teams that already run those on-premises.
A private registry inside the network, such as Red Hat Quay, Harbor, or Artifactory, holds mirrored OpenShift, IBM, and dependency images. An ImageDigestMirrorSet redirects image pulls from public registries to that mirror.
MAS integrates with corporate identity through SAML or LDAP, and OpenShift routes expose the suite on internal DNS. Certificates come from cert-manager, which can use an internal certificate authority.
Trade-offs
30 Sep 2025 maximo 7.6.1.x base support ended. Regular base support for Maximo Asset Management 7.6.1.x ended; Extended and Sustained Support are optional purchases
18 months openShift maintenance per minor release. Included with the OpenShift subscription that comes with a MAS license, plus an Extended Update Support add-on term
500 GB+ minimum private registry storage. Minimum capacity the MAS CLI asks for when it sets up an in-cluster registry for a disconnected install, with 2,000 GB recommended
Teams that ran Maximo 7.6 on WebSphere now operate a Kubernetes platform with operators, catalogs, and cluster upgrades. That calls for different skills and runbooks than application server administration.
The vSphere CSI driver only provides ReadWriteMany volumes when vSAN file service is configured. Without it, a ReadWriteMany request fails, so many sites add OpenShift Data Foundation or NFS, each with its own sizing and support model.
A disconnected install depends on mirrored images for OpenShift, IBM operators, the MAS applications, Db2, and MongoDB. The MAS CLI's own registry setup asks for at least 500 GB, and every catalog update needs a fresh mirror before it can be applied.
Each IBM Maximo operator catalog supports a defined set of OpenShift releases, and each OpenShift minor release has its own support window. Clusters behind a mirror can drift out of support if updates are not scheduled.
In-cluster Db2 and MongoDB keep data inside the data center, but backup, restore, and disaster recovery become the platform team's job rather than a managed service's.
Moving to MAS 9 changes the runtime, authentication, and integration endpoints. Customizations, Java classes, and reports built for 7.6 need to be reviewed before the database is upgraded into Maximo Manage.
What we do
We review your vSphere capacity, storage, network segmentation, identity provider, and database standards against MAS 9 requirements, and confirm OpenShift and catalog versions with IBM's compatibility reports for the release you plan to run.
AI on On-Premises
The assessment covers whether Maximo Visual Inspection or the MAS AI Service is in scope, since GPU worker nodes and extra capacity change the cluster design.
We install OpenShift on vSphere with the method that fits your environment, set up the ReadWriteOnce and ReadWriteMany storage classes, and configure ingress, certificates, and identity before MAS goes in.
AI on On-Premises
Red Hat supports NVIDIA GPUs on vSphere through GPU passthrough or NVIDIA vGPU, with the NVIDIA GPU Operator managing drivers in the cluster, so Visual Inspection can run on the same platform.
For OT and plant networks, regulated sites, and data residency requirements, we mirror images with mas mirror-images and mas mirror-redhat-images, either directly or through a two-phase transfer to removable media, then run mas configure-airgap and install from the private registry.
AI on On-Premises
Visual Inspection and AI Service images are mirrored with the rest of the suite, so inspection models can run on plant data that never leaves the site network.
We inventory customizations, Java extensions, integrations, and reports, rehearse the database upgrade into Maximo Manage on a copy of production, and plan the cutover so your existing on-premises database platform can be reused where supported.
AI on On-Premises
Once on MAS 9, Manage data can feed Maximo Health, Predict, and the Maximo Assistant without a separate integration project.
We set up runbooks for catalog updates, OpenShift minor upgrades, Db2 and MongoDB backups, and mirror refreshes, so disconnected clusters stay on supported versions.
AI on On-Premises
GPU drivers and the NVIDIA GPU Operator are included in the upgrade runbook so AI workloads keep running after OpenShift updates.
Walk through your cluster, databases, integrations, and AI plans with an engineer who installs and runs MAS on OpenShift.
The stack
Requirements
IBM's guide to installing MAS in a disconnected environment with a private mirror registry, a static catalog, and the automatic approval strategy.
Install preparation, ReadWriteOnce and ReadWriteMany storage class selection, and in-cluster or bring-your-own Db2 and MongoDB.
mas mirror-images modes (direct, to-filesystem, from-filesystem) and the content stages for core, applications, Db2, and MongoDB.
mas configure-airgap creates the registry pull secret, the ImageDigestMirrorSet, and the CA trust the cluster needs to pull from a private registry.
Installation methods for vSphere, including installer-provisioned, user-provisioned, Agent-based, and Assisted installs and restricted network options.
A small Quay-based registry for disconnected installs. Red Hat states it is not a substitute for a production Red Hat Quay deployment.
The thin-csi storage class and the vSAN file service requirement for ReadWriteMany volumes on vSphere.
Deploying ODF to provide block and shared file storage classes on vSphere-based OpenShift clusters.
Yes. MAS 9 runs on Red Hat OpenShift, and OpenShift can run on VMware vSphere, other virtualization, or bare metal in your own data center. A MAS license includes an OpenShift subscription, so the main decisions are infrastructure, storage, and databases.
MAS runs on OpenShift, and Red Hat supports OpenShift on vSphere with installer-provisioned, user-provisioned, Agent-based, and Assisted install methods. Check IBM's compatibility reports for the exact OpenShift versions supported by your MAS catalog.
Yes. IBM documents disconnected installs using a private mirror registry and a static operator catalog. The MAS CLI mirrors images with mas mirror-images, configures the cluster with mas configure-airgap, and the installer detects the disconnected setup from the ImageDigestMirrorSet it creates.
MAS needs a ReadWriteOnce and a ReadWriteMany storage class. The vSphere CSI driver's thin-csi class supports ReadWriteMany only with vSAN file service configured, so many sites use OpenShift Data Foundation, which provides both classes, or NFS.
Yes. The MAS CLI can provision Db2 and MongoDB in the cluster, or you can bring your own databases. Maximo Manage also supports Oracle Database and Microsoft SQL Server as alternatives to Db2.
It is a small Quay-based registry Red Hat provides for hosting the images needed to install a disconnected OpenShift cluster. Red Hat says it is not a substitute for a production Red Hat Quay deployment, so long-running MAS environments usually mirror into an enterprise registry such as Quay, Harbor, or Artifactory.
IBM ended regular base support for Maximo Asset Management 7.6.1.x on 30 September 2025. Eligible customers on 7.6.1.3 or later can buy Extended or Sustained Support, but IBM's long-term path is MAS 9.
Yes. The Maximo 7.6 database is upgraded into Maximo Manage on an OpenShift cluster, which can sit in the same data center. Customizations, integrations, and reports need review first, and a rehearsal on a copy of production is standard practice.
Red Hat supports NVIDIA GPUs for OpenShift on vSphere through GPU passthrough or NVIDIA vGPU, with the NVIDIA GPU Operator managing drivers. The MAS automation uses the certified GPU Operator when it sets up GPU support.
Common reasons are OT and plant networks that cannot reach the internet, data residency rules, regulated environments, and existing data center capacity. The trade-off is that your team operates OpenShift, storage, and databases instead of a managed service.
Tell us where Maximo runs today and where you want it to run. We will tell you honestly whether On-Premises is the right fit.
