+ 
+ 
Edge orchestration tools like Avassa place containers on remote hosts. Industrial operations need more than that: equipment data, local control, an immutable OS, zero-touch provisioning and GitOps across hundreds of sites. SUSE Edge, Losant (now SUSE Edge Operational Intelligence) and Krumware deliver all of it on open, Kubernetes-native foundations.
An edge orchestrator answers "where does this container run?" Operators also have to answer how to get data out of 20-year-old PLCs, what happens when the satellite link drops, who installs and patches the OS on 300 boxes, and how site two avoids becoming a one-off science project.
PLCs, meters and drives speak Modbus, OPC UA, S7 and BACnet. Someone has to connect, normalize and act on that data, ideally without writing a custom gateway per site.
Satellite, cellular and congested WANs mean sites must make decisions locally, buffer data and reconcile on their own when they come back.
Provisioning, OS patching, rollback and app delivery across hundreds of sites need declarative, Git-driven automation, not SSH and spreadsheets.
SUSE acquired Losant in February 2026. The industrial IoT platform now sits on top of SUSE Edge as SUSE Edge Operational Intelligence, and Krumware engineers the whole thing into a working platform for your sites.
Protocol connectors, edge and embedded workflows, dashboards, digital twins and white-label apps.
One console and one Git workflow for every cluster, from the data center to the smallest site.
CNCF-certified Kubernetes sized for the edge, with zero-trust runtime security, replicated storage and VMs for legacy workloads.
SUSE Linux Micro with transactional updates and rollback. Elemental and Edge Image Builder take a box from power-on to cluster with zero touch.
A hypothetical operator runs compressors, pumps and meters across remote sites connected by satellite and cellular. Press play, or jump to any step, then tap sites to cut their links yourself.
Each site runs the Losant Gateway Edge Agent (SUSE Edge Operational Intelligence) as a container on SUSE Edge. It reads pumps, compressors and meters natively over OPC UA, Modbus TCP and MQTT. There is no protocol gateway to build or maintain.
Edge workflows normalize, downsample and threshold thousands of tags locally. Only changes, aggregates and events cross the satellite or cellular link, cutting uplink volume by roughly 95% in this scenario.
A vibration spike on a compressor trips an edge workflow, which sends a shutdown write back to the PLC. The decision is made on site with no cloud round-trip. Central dashboards learn about it on the next sync.
A merged pull request is the release. Fleet picks it up in Rancher and promotes it to two canary sites, checks health, then rolls through the fleet in waves. Krumware builds and runs the pipeline and the repo layout behind it.
Five sites drop off the network for hours. Their clusters and edge workflows keep operating, local alarms still fire, and the agent buffers telemetry to disk. The control plane marks them stale, not down.
When links return, Fleet reconciles the sites to the desired state, so the release they missed installs on its own. Buffered telemetry back-fills with no gaps in history. Nobody has to visit.
A new industrial PC ships with an image built by Edge Image Builder. On first boot it registers with Elemental, comes up as a K3s/RKE2 cluster on immutable SUSE Linux Micro, receives its Fleet bundles and starts streaming data, all the same day.
Tap any site to cut or restore its link.
Hypothetical scenario for illustration. Site counts, rates and timings are representative, not measured.
A Losant edge workflow watches a pump's vibration over Modbus and acts in milliseconds. Drag the trip limit, simulate a bearing fault, then cut the uplink to see store-and-forward in action.
Cut the uplink and trip the pump. The decision still happens on site, and the events back-fill when the link returns.
Every connection starts at the site and goes outbound. Pick a flow to trace it through the stack.
PLCs and sensors → Gateway Edge Agent (OPC UA, Modbus, MQTT) → edge workflows → outbound MQTT → SEOI control plane → dashboards, digital twins, ERP/CMMS.
Avassa is a capable, edge-native application orchestrator, and on rollouts, offline autonomy, secrets and VMs it is genuinely comparable. The difference is everything above and below that layer.
Avassa Runs the containers you bring. No dashboard or IIoT app builder is described on its platform pages.
Sources [1] docs.losant.com[2] avassa.io
Avassa Application logic is yours to write and package. Avassa positions as app-centric rather than a "data forwarding and connectivity" platform.
Avassa No OT protocols are named on the industrial page. Protocol handling comes from containers you supply.
Sources [1] docs.losant.com[2] avassa.io
Avassa The Edge Enforcer needs a Linux host with a container runtime such as Docker or Podman.
Sources [1] docs.losant.com[2] avassa.io
Avassa A core strength: label-based placement with rolling and canary releases from the Control Tower.
Avassa Also strong: site-local clustering, state replication and an offline secrets vault.
Avassa Also strong: built-in secrets distribution and encrypted application networking.
Sources [1] suse.com[2] avassa.io
Avassa The Edge Enforcer can schedule VMs as well as containers.
Avassa "Avassa does not require Kubernetes." It uses its own application specification; existing Kubernetes apps are converted.
Sources [1] avassa.io
Avassa Runs on a Linux you provide (Debian/Ubuntu, Red Hat, Wind River, Yocto…). It can roll out OS upgrades but does not supply the OS.
Sources [1] suse.com[2] avassa.io
Avassa Initial OS install and bare-metal provisioning are not described; hardware and OS come via partners.
Avassa No open-source licensing is claimed. Avassa's own comparison credits open source to Kubernetes only.
Sources [1] suse.com[2] avassa.io
Avassa Vendor support plus partners such as Elastisys and hardware/OS vendors.
Sources [1] krum.io[2] avassa.io
Based on each vendor's public documentation as of November 2026. "Bring your own" and "via partners" mean the capability is not described in Avassa's own product pages. Tap a row for detail and sources. Corrections welcome.
you only need to place and update containers on a few Linux hosts per site, and you already have the OS, device provisioning, equipment connectivity and operator apps covered elsewhere.
you want one supported path from bare metal to dashboards: immutable OS, zero-touch provisioning, CNCF Kubernetes, native OT protocols, edge workflows and a partner who builds it with you.
Avassa positions against Kubernetes and converts apps into its own spec. SUSE Edge keeps the same Kubernetes API, Helm charts and GitOps tooling you use in the data center, so there's one skill set end to end.
Krumware is a platform engineering firm, a long-time Rancher partner, SUSE's 2024 Emerging Technology Solutions Partner of the Year, and the maintainer of Epinio. Its edge practice starts from SUSE Edge and adds the engineering that turns products into a platform.
Sites, equipment, protocols, network and compliance constraints, mapped to a target architecture and rollout plan.
Rancher Accelerate for Edge: send a device, and Krumware prepares the image, prototypes your workload, stands up the platform and trains your team.
A handful of real sites with Losant edge agents, GitOps delivery, observability and runbooks, measured against agreed outcomes.
Waves of sites with zero-touch onboarding, air-gap packaging where needed, and enterprise support from Krumware and SUSE.
Tell us about the equipment, the network and the constraints. We'll show you the stack running against them.
+